ARK analysts project SpaceX-Tesla merger, $1T SpaceX revenue, and Tesla's Robotaxi dominance
Summary
ARK Invest analysts discuss Elon Musk's $1 trillion revenue guidance for SpaceX, noting it is five times higher than consensus estimates, driven by aggressive terrestrial and orbital compute expansion. The panel assigns a 90%+ probability to a SpaceX-Tesla merger within 18 months, citing operational integration but highlighting regulatory and shareholder hurdles. For Tesla, analysts attribute 80% of enterprise value to the Robotaxi business, forecasting $60 billion in EBIT by 2030, while Optimus is valued at 14% with a forecast of 1 million units by 2031. Key risks identified include debt market skepticism, geopolitical tensions affecting the merger, and execution delays in autonomy scaling.
Key takeaways
- 1 Consensus 2028 revenue estimates for SpaceX were around $200-220 billion, making Elon Musk's guidance of $1 trillion roughly 5x higher than analyst expectations, forcing a complete model rewrite.
Unidentified speaker (speaker_3) · 3:20
- 2 Delivering a gigawatt of compute in orbit via Starlink infrastructure will cost approximately one-third of what ground-based competitors charge, creating a structural cost advantage for orbital AI.
Unidentified speaker (speaker_2) · 31:29
Episode sections
- 0:00 – 2:36
Intro and Sponsor Read
Speaker 0 introduces the podcast and the guests, followed by a standard ARK Invest disclaimer and sponsor read which is ignored per instructions.
- 2:36 – 5:25
SpaceX Revenue Guidance and Analyst Reaction
Alex Potter discusses Elon Musk's $1 trillion revenue guidance for SpaceX, noting it is roughly 5x higher than the 2028 consensus of $200-220 billion. He describes the reaction as 'absurd' and notes the need to rebuild models from scratch.
- 5:25 – 7:16
Terrestrial Compute Capacity (20 GW)
Brett Winton identifies 20 gigawatts of power and cooling capacity for terrestrial compute as a key metric. He contrasts this with previous guidance of 5-10 GW, highlighting the accelerated pace of SpaceX's build-out compared to typical data center constraints.
- 7:16 – 9:34
Starship Launch Cadence
Aaron Burnett highlights a target of 1-2 Starship launches per week. He argues this cadence supports a massive increase in Starlink bandwidth (40-60 launches next year) and is foundational for orbital compute projects like Starmind.
- 9:34 – 18:23
AI Strategy: Grok, Cursor, and Compute Pricing
The group discusses SpaceX's position in the 'Frontier League' via Grok and the Cursor acquisition. Brett Winton asserts that compute pricing will remain high ($50B+/GW) due to insatiable demand, allowing SpaceX to monetize infrastructure at premium rates.
- 18:23 – 25:58
Funding the Build-Out
Pierre Ferragu and Brett Winton discuss how SpaceX will fund the $50B per GW infrastructure. Ferragu compares SpaceX to a 'surfer' riding the wave of demand, noting that high pricing allows the CapEx to self-fund through debt against assets, potentially avoiding equity dilution.
- 25:58 – 27:31
Starlink Enterprise and Connectivity
Tasha Keeney notes the shift from consumer to enterprise/government focus for Starlink, citing spectrum acquisitions and a total connectivity market opportunity of ~$1 trillion. Revenue is tied to bandwidth capacity enabled by Starship launches.
- 27:31 – 28:12
Direct-to-Cellular Competition
Aaron Burnett discusses SpaceX's direct-to-cellular strategy, suggesting it will be head-to-head with terrestrial carriers like T-Mobile and Verizon, though the execution method may differ.
- 28:12 – 33:58
Starlink Connectivity, Terrestrial Compute, and Orbital Compute Economics
Speaker 3 discusses how Starlink's satellite architecture bypasses traditional telco spectrum constraints, targeting 60% of the US population density for V3 satellites. The group debates capital allocation between terrestrial data centers, direct-to-cell satellites, and orbital compute, with Speaker 2 arguing that orbital infrastructure costs for a gigawatt of compute are structurally lower than ground-based alternatives.
- 33:58 – 39:09
Starship Flight Test 15 and Launch Cadence Targets
Speaker 3 predicts success for Starship Flight Test 15, emphasizing that catching the upper stage with 'chopsticks' is critical for rapid reusability. Speaker 3 notes that while industry experts doubt the 1-2 launches per week target, SpaceX is currently building one rocket a month, making the ramp to 50+ annual launches feasible and potentially quadrupling orbital bandwidth.
- 39:09 – 48:31
SpaceX and Tesla Merger Probability and Shareholder Dynamics
Brett Winton (Speaker 2) and Pierre Ferragu (Speaker 3) assign a 90%+ probability to a SpaceX-Tesla merger within 18 months, citing deep operational integration like Grok in cars and Terrafab. Alex Potter (Speaker 1) agrees it is more likely than not but highlights regulatory hurdles in China and the challenge of convincing Tesla shareholders to accept a premium, given Elon Musk's valuation expectations.
- 48:31 – 59:05
Market Skepticism, Debt Markets, and Execution Risks
Cathie Wood (Speaker 3) identifies debt market risk aversion and 'scar tissue' from the 1990s as a major downside risk, comparing current growth to the fiber boom. Alex Potter cites potential financing freezes, while Brett Winton argues that market skepticism regarding AI compute demand (Anthropic/OpenAI) creates a mispricing opportunity, though he acknowledges execution risks on aggressive schedules.
- 59:05 – 59:34
Tesla Valuation Model and RoboTaxi Dominance
Tasha Keeney (Speaker 1) introduces ARK's latest Tesla valuation model, stating that approximately 80% of the company's enterprise value is derived from the RoboTaxi business, consistent with previous models.
- 59:34 – 1:35:27
Tesla Valuation Components and Optimus Forecast
Daniel Maguire outlines Tesla's enterprise value drivers, attributing roughly 80% to Robotaxi and 14% to Optimus, with a forecast of 1 million Optimus units deployed by 2031. He also highlights a projected $225 billion cumulative spend on AI compute by 2031 to enhance Optimus capabilities.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| TeslaTSLA | equity | mixed | On October 26th, we had an amazing live broadcast of what we dubbed the aftercall call. And this was the call following SpaceX and Tesla's earnings calls that we hosted in collaboration with Mach |
| SpaceX | private company | mixed | Um I think like Brett really brings the answer to your question, which is lah I like to compare SpaceX in the data center build out as a surfer at the foot of the wave, being pushed by the wave, and having actually a lot of flexibility and maneuverability. So today uh SpaceX can sell for about $50 billion dollars a year, a data center that as you said costs $50 billion dollars to deploy. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
SpaceX 2028 revenue consensus was $200-220 billion, making Elon Musk's guidance roughly 5x wrong.
Unidentified speaker (UNKNOWN) ·3:19
How to verify: Check Piper Sandler research reports or consensus estimates for SpaceX revenue for the year 2026 prior to the earnings call. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX has lined up 20 gigawatts of power and cooling capacity for terrestrial compute.
Unidentified speaker (speaker_2) ·5:33
How to verify: Review SpaceX or X Corp earnings call transcripts or official announcements regarding data center infrastructure plans. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX previously guided to 5-10 GW of compute capacity by the end of 2027.
Unidentified speaker (speaker_2) ·5:53
How to verify: Review prior SpaceX or X Corp investor presentations or earnings call transcripts for 2026 guidance. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX is targeting 1-2 Starship launches per week.
Unidentified speaker (UNKNOWN) ·7:32
How to verify: Monitor SpaceX launch schedules and official Starship flight logs for weekly cadence. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX aims for 40-60 Starship launches next year.
Unidentified speaker (speaker_3) ·8:45
How to verify: Compare SpaceX launch manifest against total annual launch capacity and historical Starship flight rates. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX can charge $50+ billion per gigawatt for compute.
Unidentified speaker (speaker_2) ·13:05
How to verify: Review SpaceX or X Corp financial disclosures or contracts with hyperscalers for infrastructure-as-a-service pricing. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Top four hyperscalers will generate $6 trillion in revenue over the next five years.
Unidentified speaker (UNKNOWN) ·19:32
How to verify: Check sell-side analyst reports from firms like New Street Research or Piper Sandler on hyperscaler revenue forecasts. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Starlink acquired low-band spectrum necessary for mobile.
Unidentified speaker (speaker_1) ·26:52
How to verify: Review FCC filings or SpaceX/X Corp announcements regarding spectrum acquisitions for mobile services. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Primary sourceCursor captures about 25% of AI tokens working on software development.
Unidentified speaker (speaker_3) ·11:06
How to verify: Check market share data from AI coding tool analytics firms or Cursor's own public metrics on token usage. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Starlink has approximately 80 megahertz of spectrum, which speakers argue is sufficient to serve 60% of the US population via satellites, bypassing the need for traditional telco spectrum licenses.
Unidentified speaker (speaker_3) ·28:12
How to verify: Check SpaceX/Starlink spectrum filings with the FCC (Federal Communications Commission) and compare against population density maps of the US. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Primary sourceSpaceX is currently building approximately one Starship rocket per month.
Unidentified speaker (speaker_3) ·36:10
How to verify: Review SpaceX production updates, Starbase facility imagery, or third-party supply chain reports from late 2026 to early 2026. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Starlink revenue is projected to clear around $500 billion in the early 2030s.
Unidentified speaker (speaker_2) ·32:11
How to verify: Check ARK Invest's published financial models or Piper Sandler/New Street Research equity research reports for Starlink revenue forecasts. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Orbital compute infrastructure costs are structurally lower, approximately one-third the cost of ground-based compute.
Unidentified speaker (speaker_2) ·31:29
How to verify: Compare SpaceX/ARK internal cost models against public data center power/cooling costs (e.g., from AWS or Azure pricing) and orbital launch/operation cost estimates. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Starship Flight Test 15 aims to catch the upper stage with 'chopsticks'.
Unidentified speaker (UNKNOWN) ·34:00
How to verify: Monitor SpaceX official announcements or live coverage of Starship Flight Test 15 scheduled for late 2026. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Launch costs could drop to $10 per kilogram at massive scale with full reusability.
Unidentified speaker (UNKNOWN) ·38:17
How to verify: Review Elon Musk's historical statements on Starship cost targets and compare against SpaceX's manufacturing cost disclosures. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
There is a 90%+ probability of a SpaceX-Tesla merger within 18 months.
Unidentified speaker (speaker_2) ·39:23
How to verify: Track SEC filings for any merger proposals or regulatory approvals involving Tesla and SpaceX; monitor public statements from Elon Musk. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Primary sourceTesla's enterprise value is approximately 80% derived from the RoboTaxi business.
Unidentified speaker (speaker_1) ·59:31
How to verify: Review ARK Invest's latest Tesla valuation model published in late 2026. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX guided to a $1 trillion revenue run rate by the end of 2027 or 2028.
Unidentified speaker (speaker_2) ·53:27
How to verify: Check transcripts of SpaceX earnings calls or Elon Musk's public statements regarding revenue guidance. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
SpaceX is 10 years ahead of competitors in achieving full reusability.
Unidentified speaker (UNKNOWN) ·37:47
How to verify: Compare SpaceX's Starship reusability milestones against Blue Origin's New Glenn and China's Long March programs. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Elon Musk put a $1 trillion revenue number on SpaceX.
Unidentified speaker (speaker_1) ·59:34
How to verify: Check SpaceX press releases or Elon Musk's public statements on X (formerly Twitter) regarding SpaceX valuation and revenue targets. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Tesla expects to have 8 million cars in the fleet by 2031.
Unidentified speaker (UNKNOWN) ·59:36
How to verify: Verify against Tesla's investor presentations or earnings call transcripts from 2026-08-05 to 2026-11-05 for fleet projections. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Tesla will deploy 1 million Optimus robots by 2031.
Unidentified speaker (speaker_1) ·59:45
How to verify: Check Tesla's Q3 2026 earnings transcript or investor day materials for Optimus production forecasts. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
FSD subscriptions have replaced 100% margin credit revenue.
Unidentified speaker (speaker_2) ·1:01:12
How to verify: Analyze Tesla's GAAP financial statements (Form 10-Q/10-K) to compare service revenue trends with automotive regulatory credit revenue. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Primary sourceTesla achieved 2 million driverless miles in eight months.
Unidentified speaker (UNKNOWN) ·1:02:15
How to verify: Cross-reference Tesla's autonomy update reports or SEC filings detailing autonomous driving data collection metrics. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Primary sourceElon Musk is spending four days a week in Memphis to ensure superintelligence gets built out for SpaceX.
Unidentified speaker (speaker_2) ·1:31:28
How to verify: Check Elon Musk's public schedule, social media posts, or SpaceX/Tesla press releases regarding his location and time allocation during the relevant period. For recent events, verify 2026-08-05 through 2026-11-05. Use another period only when the speaker explicitly names it.
Do your own research
- Verify SpaceX's 2028 revenue consensus estimates from major investment banks (e.g., Bloomberg Terminal, Refinitiv) to confirm the $200-220 billion baseline.
- Check SpaceX's actual Starship launch cadence and production rates against the stated goal of 1-2 launches per week via public flight logs.
- Review Tesla's Q3/Q4 earnings reports for FSD subscription revenue trends and unit counts to assess the shift from credit revenue to high-margin subscriptions.
- Compare Tesla's estimated cost per mile for CyberCab against Waymo's published operational costs to validate the 40% cost advantage claim.
- Assess regulatory filings (SEC, DOJ) for any updates on potential antitrust or geopolitical barriers to a SpaceX-Tesla merger.
- Examine Tesla's driverless mileage data releases to verify the claim of 2 million miles in eight months versus Waymo's deployment rate.
- Analyze ARK Invest's specific valuation models for Tesla, particularly the assumptions behind the 80% Robotaxi and 14% Optimus value attribution.
- Check for any conflicts of interest or forward-looking statement disclaimers in ARK Invest's published research materials regarding these projections.
Full transcriptexpand
- 0:00
- Unidentified speaker (speaker_0): Welcome to FYI, the four-year innovation podcast. On October 26th, we had an amazing live broadcast of what we dubbed the aftercall call. And this was the call following SpaceX and Tesla's earnings calls that we hosted in collaboration with Mach
- 0:20
- Unidentified speaker (speaker_0): 33 to discuss everything that happened on those calls as well as what we can look forward to in the future. We said, wouldn't it be great if we had the sell side and the buy side and deep research all come together? And we were honored to have Alex Potter from Piper Sandler and Pierre Farragut from New Street Research join us.
- 0:40
- Unidentified speaker (speaker_0): And we hope you enjoy the conversation.
- 0:43
- Unidentified speaker (speaker_1): Welcome to FYI, the four year innovation podcast. This show offers an intellectual discussion on technologically enabled. disruption. Because investing in innovation starts with understanding it. To learn more, visit Arc-Invest.com.
- 1:01
- Unidentified speaker (speaker_2): Arc Invest is a registered investment advisor focused on investing in disruptive innovation. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. It does not constitute either explicitly or implicitly any. provision of services or products by ARC. All statements made regarding companies or securities are strictly beliefs and points of view held by
- 1:21
- Unidentified speaker (speaker_2): ARC or podcast guests and are not endorsements or recommendations by ARC to buy, sell, or hold any security. Clients of ARC Investment Management may maintain positions in the securities discussed in this podcast.
- 1:37
- Unidentified speaker (speaker_0): Welcome to the Call After the Calls. Excited for everyone to join us. We've got amazing guests talking about SpaceX and Tesla. Of course, we've got the entire ARC team here. We've got Tasha, we've got Daniel, we've got Kathy, we've got Brett, and we've got some amazing guests as well
- 1:57
- Unidentified speaker (speaker_0): externally. Don't forget Daniel there as well. We've got Pierre Farragou Farrago of New Street Research, Sean McGuire. joining later from Sequoia, Aaron Burnett, of course, of Mach 33, and Alex Potter from Piper Sandler. So an amazing combination: sell side, research, buy side. We'll be able to get all
- 2:16
- Unidentified speaker (speaker_0): perspectives here on all things SpaceX and Tesla. We're live across X, LinkedIn, YouTube, TikTok. I think you can find us anywhere. So definitely drop comments and questions. We've got people monitoring all of those lines. We'd love to be able to answer some insightful questions. All right, we're gonna kick
- 2:36
- Unidentified speaker (speaker_0): this off with the number that mattered. So let's pull in Alex Potter from Piper, Brett from Arc, and Aaron Burnett from Mach Mock 33. thirty three. Alex, let's start with you. The number the number that you called out, one trillion dollars in revenue.
- 2:56
- Unidentified speaker (speaker_0): Why is that the number that matters to you?
- 3:01
- Unidentified speaker (speaker_3): Uh geez, I don't know. I mean, because it made me dizzy and almost faint when I heard it. Um
- 3:08
- Unidentified speaker (UNKNOWN): I
- 3:08
- Unidentified speaker (speaker_3): don't know. I I've never I've never covered a company where it's certainly not a company that's this liquid and high profile,
- 3:18
- Unidentified speaker (speaker_3): where
- 3:19
- Unidentified speaker (UNKNOWN): yeah,
- 3:20
- Unidentified speaker (speaker_3): I don't know how many cell side firms cover SpaceX. It's 50 or something, 60 different firms that cover it. And I think consensus in 2028 was like 200 billion, 220 20 billion, something like that. So roughly 5x wrong. Uh relative
- 3:39
- Unidentified speaker (speaker_3): to what Elon said uh 2028 revenue could be. That's absurd. Um and like I said, I like it's you after the call is done as a as an analyst, you sit there staring at your model,
- 3:55
- Unidentified speaker (speaker_3): not really knowing what to do. I've never, it's it's not like other calls, it's not like other companies where you're like, oh, you it know, guidance. needs to be we need to tweak our numbers a little bit. This is like, no, we need to take our model down to the studs. Um, it's it's a fun company to cover in that sense, but it's also it's
- 4:15
- Unidentified speaker (speaker_3): just it's unreal.
- 4:17
- Unidentified speaker (speaker_0): So so is that what you've been doing this past week? How big of a rewrite has the underwriting team done?
- 4:25
- Unidentified speaker (speaker_3): Yeah, I mean, it's ongoing.
- 4:28
- Unidentified speaker (speaker_3): And I know that Aaron does a much better job of of you know modeling uh Tesla than than I do. You know, I've covered or Tesla, SpaceX. I I always do that.
- 4:41
- Unidentified speaker (speaker_3): So
- 4:42
- Unidentified speaker (speaker_3): I've covered I've covered Tesla for a really long time. So like that the it just always slips out. Um, And our Tesla model is extremely detailed, right? It took us a long long time to build it. And I like talking about companies with that level of granularity. And we haven't with SpaceX. And so, and my SpaceX model is still more detailed than the average model of
- 5:02
- Unidentified speaker (speaker_3): the companies that I have under coverage. But when Elon comes out and says things like that, and it sort of illuminates how wrong the model really is, it it makes you sit. there and stare at your screen. And I and that's basically I've either literally or figuratively been doing that basically ever since the call last week.
- 5:22
- Unidentified speaker (UNKNOWN): Uh
- 5:23
- Unidentified speaker (speaker_3): all
- 5:23
- Unidentified speaker (UNKNOWN): right,
- 5:23
- Unidentified speaker (speaker_0): Brett.
- 5:23
- Unidentified speaker (speaker_3): Just thinking through.
- 5:25
- Unidentified speaker (speaker_0): Brett, the number that you said directly ties to this. 20 gigawatts. Why
- 5:31
- Unidentified speaker (speaker_2): is
- 5:31
- Unidentified speaker (speaker_0): that the number for you?
- 5:33
- Unidentified speaker (speaker_2): Well, I think 20 gigawatts is both what they Elon said they had lined. up for um basically cooling in power capacity to for their terrestrial compute build out. And um that was the biggest change for us in terms of how we thought about the trajectory of the company,
- 5:53
- Unidentified speaker (speaker_2): in that they had previously guided to by the end of the year, you know, maybe during doing five to ten 10 gigawatts by the end of 27, five to ten 10 gigawatts installed. And I was mentally prepared for them to say, you know what, building this stuff is really hard and we're going as fast as we can. And it's going to come in around the five gigawatt range. And instead they said, well, we're going to
- 6:13
- Unidentified speaker (speaker_2): come in around 13. And there's a probability that we get to 20. Um, and so relative to the entire kind of like narrative around data centers being hard to build and there being permitting and power issues. Um, Elon really is able to drive his companies at a maniacal pace. And that 20
- 6:33
- Unidentified speaker (speaker_2): gigawatt number, or even the 13 gigawatt number, you can use to get to some of the Gaudi revenue figures that he put out there. So 500 billion, you know, um run rate revenue by the end of 27, uh leading to that trillion dollar number. Uh is kind of what falls out of the model
- 6:53
- Unidentified speaker (speaker_2): as you kind of drive forward the terrestrial business in that way.
- 6:58
- Unidentified speaker (speaker_0): I I think he was a little bit uh more more forceful when he said that he said not only did he say it's not just five gigawatts, he said we have rocket engineers and they're playing in the little leagues. So
- 7:12
- Unidentified speaker (speaker_3): that quote yeah
- 7:12
- Unidentified speaker (speaker_0): yeah it's a it
- 7:13
- Unidentified speaker (UNKNOWN): was
- 7:13
- Unidentified speaker (speaker_0): a little bit
- 7:14
- Unidentified speaker (UNKNOWN): Yankees
- 7:14
- Unidentified speaker (speaker_3): against the
- 7:15
- Unidentified speaker (speaker_0): Exactly.
- 7:16
- Unidentified speaker (speaker_0): Uh Aaron, Aaron, your number here is uh one to two starship launches per week. Uh I like this. This is this is different from just focused on the immediate term uh data center opportunity. For you, why is that the key number here?
- 7:32
- Unidentified speaker (UNKNOWN): Well
- 7:33
- Unidentified speaker (speaker_3): I should sh I should mention right off the bat that uh Alex uh get used to looking dumb with uh SpaceX. Uh no matter what we put out there. I've
- 7:42
- Unidentified speaker (UNKNOWN): gotten used to it
- 7:43
- Unidentified speaker (speaker_2): by now, even in the
- 7:45
- Unidentified speaker (speaker_3): It is just a uh yeah, i they're move they move fast for a rocket company and now you add in the AI of it all and it's now just yeah, just everything's moving much faster. I guess this is the feeling of being in singularity now. So um yeah, so one to two weekly cadence is uh the number. And I we're focused on that because I think
- 8:05
- Unidentified speaker (speaker_3): it points to a much bigger Starlink kind of bandwidth than people are expecting. And what I found is even some of the most bullish and sharpest people I know have a visceral gut reaction that that one to two equals something like fifty 50 to a 100 hundred launches next year of Starship. And the
- 8:25
- Unidentified speaker (speaker_3): the smartest people I know are just like, uh, yeah, that doesn't feel right. It's way too high. And so, you know, I've just recently put out something today that was on this. Like, if you look at what they're doing and where they're headed, it's it's actually not too crazy. Uh, it's not like what people tend to think about for maybe Elon numbers where maybe it's a few years uh
- 8:45
- Unidentified speaker (speaker_3): early or something like that. I think we're looking at a realistic um upmass up mass of you know somewhere. in that forty to sixty range of number of launches we can see next year, which is very big. It's something like doubling or tripling the total bandwidth on orbit that Starlink can provide. And so that really that really starts to throw a wrench into people's numbers.
- 9:05
- Unidentified speaker (speaker_3): Um and then it also is the is the foundation for Starmont, um, which I think just continues is really like uh beyond this twenty 20 gigawatts it's gonna be uh
- 9:15
- Unidentified speaker (UNKNOWN): of
- 9:15
- Unidentified speaker (speaker_3): terrestrial in theory that would be the the low percent probability. But if that's you know on land in the next few years or the next year to two, um they need to what's the next hundred gigawatts or whatever that they're aiming for internally, and that's gonna come from Starmine. So you really need to see this ramp up of of Starship.
- 9:33
- Unidentified speaker (UNKNOWN): All
- 9:33
- Unidentified speaker (speaker_0): right, so what we're
- 9:34
- Unidentified speaker (UNKNOWN): gonna
- 9:34
- Unidentified speaker (speaker_0): do now is we're gonna tackle uh SpaceX from kind of each of their sections. segments here. We're gonna dive into superintelligence. We'll talk about Starlink and connectivity and Starship. Uh, but maybe as we dive into superintelligence, we'll bring Pierre on screen. We booked Pierre from New Street Research via his
- 9:53
- Unidentified speaker (speaker_0): crockbot. And Elon on the call saying Grokbot, the fastest growing product that he has ever seen. How important do you think that is? to the future of SpaceX?
- 10:08
- Unidentified speaker (speaker_3): Well, it's definitely um
- 10:11
- Unidentified speaker (UNKNOWN): well,
- 10:11
- Unidentified speaker (speaker_3): I would say it's um SpaceX is in a very unique position in the in the Frontier League. So they are in the Frontier League. Groc is uh really qualified as
- 10:20
- Unidentified speaker (UNKNOWN): a
- 10:20
- Unidentified speaker (speaker_3): frontier model, even if it's uh if it's still uh uh mostly uh mostly running behind. And it's very important for the long-term future of SpaceX to
- 10:30
- Unidentified speaker (speaker_3): um in that league. And um over the last six months we went from having a model
- 10:36
- Unidentified speaker (UNKNOWN): that
- 10:37
- Unidentified speaker (speaker_3): could do well as a chat box, uh that could qualify for the for the frontier to having um like the most talked about
- 10:46
- Unidentified speaker (UNKNOWN): uh
- 10:47
- Unidentified speaker (speaker_3): eight A uh but and then the cursor acquisition.
- 10:51
- Unidentified speaker (speaker_3): is um is like um a mastermind move to um
- 10:55
- Unidentified speaker (UNKNOWN): from
- 10:56
- Unidentified speaker (speaker_3): the perspective of SpaceX, and I guess it played a lot in the development of the of the of the butt as well. Um cursor I think captures about 25
- 11:06
- Unidentified speaker (speaker_3): uh of AI uh tokens um um uh working on software development. It's uh it's it actually um uh put puts uh space in the deep into the league with um uh a a very significant market um and so you combine together grok the computer and uh
- 11:25
- Unidentified speaker (UNKNOWN): uh uh
- 11:26
- Unidentified speaker (speaker_3): and cursor and um all the ingredients are there for uh for space to be uh to be a leader
- 11:33
- Unidentified speaker (UNKNOWN): and
- 11:33
- Unidentified speaker (speaker_3): and maybe the last element elements I would want to say on that front is
- 11:38
- Unidentified speaker (UNKNOWN): how
- 11:38
- Unidentified speaker (speaker_3): well the company understands, how well Elon understands the fact that you
- 11:42
- Unidentified speaker (UNKNOWN): you
- 11:42
- Unidentified speaker (speaker_3): don't have to worry too much about articulating
- 11:45
- Unidentified speaker (UNKNOWN): your
- 11:46
- Unidentified speaker (speaker_3): leading edge model, your harness, uh, your agent, um with your compute.
- 11:57
- Unidentified speaker (speaker_3): And as long as you grow compute faster than everybody else, You uh you win. Uh when one uh way or another. And that's exactly what's uh what's happening at the moment.
- 12:06
- Unidentified speaker (UNKNOWN): So,
- 12:07
- Unidentified speaker (speaker_0): Brett, one of the big assertions that was made on the call is
- 12:10
- Unidentified speaker (UNKNOWN): that
- 12:11
- Unidentified speaker (speaker_0): the price for compute that people are willing to pay is going to keep going up.
- 12:17
- Unidentified speaker (speaker_0): Is that true?
- 12:19
- Unidentified speaker (UNKNOWN): Do
- 12:19
- Unidentified speaker (speaker_0): you believe it to be true? And uh and then we'll we'll go to the sell side and see see what they think.
- 12:26
- Unidentified speaker (UNKNOWN): Well
- 12:26
- Unidentified speaker (speaker_2): it's certainly true in the immediate term. And I think of all companies, SpaceX is going to test the limits of how much compute uh the market can absorb in a very short time. And because they have they're uniquely positioned because they have Grokbot and the cursor business, where if they are
- 12:45
- Unidentified speaker (speaker_2): over-serving the market, uh as in if demand for uh immediate immediate-term term compute. begins to soften, they can take it internally and then train to compete with that frontier. I think if you look at Anthropics S1, it's really clear why they're willing to spend a lot for compute because their financial metrics are
- 13:05
- Unidentified speaker (speaker_2): just ginormously outstanding. And um, so that they are willing to you know, rent compute at the margin for 50 plus billion per gigawatt. um, or that the market is, is because that this is mana from heaven in terms of cash flow that's sitting out there addressing an enterprise market that we think is a $40 trillion market uh
- 13:25
- Unidentified speaker (speaker_2): at full penetration. Um so um we think that actually there's plenty of uh revenue to absorb the compute that SpaceX pushes pushes into the market. And specific to SpaceX, maybe unique of all companies, they will want to and should want to test the limits of how much compute
- 13:45
- Unidentified speaker (speaker_2): the market can absorb, because then they will begin to absorb the compute internally to power Grok to kind of drive their revenue. And this chart shows what we think is going to be their They're recognized monetization over time. As you can see in our model, we're under where they're saying they're at. As in, this is 40, 30 to 40
- 14:05
- Unidentified speaker (speaker_2): billion per gigawatt that SpaceX will begin to monetize at. Whereas NeoClouds that are signing longer term contracts really don't get above 20 billion per gigawatt until the early 2030s. And so by charging at the margin for the scarce supply, SpaceX is able to recognize a lot more revenue just on an infrastructure as a service side and potentially
- 14:25
- Unidentified speaker (speaker_2): power a software business that's even more. uh margin accretive.
- 14:29
- Unidentified speaker (speaker_0): Alex, are you thinking about this in a similar way? And
- 14:33
- Unidentified speaker (UNKNOWN): you
- 14:34
- Unidentified speaker (speaker_0): know he also is asserting that there's going to be a number of stranded assets. And so, you know, you really should double down on the players who can execute. So from your perspective,
- 14:47
- Unidentified speaker (UNKNOWN): you
- 14:47
- Unidentified speaker (speaker_0): know, you're cut you're not just covering SpaceX, you're looking at other players as well. How does the compute game shake out here?
- 14:56
- Unidentified speaker (speaker_1): Yeah, I mean, that is um,
- 14:59
- Unidentified speaker (UNKNOWN): you know,
- 14:59
- Unidentified speaker (speaker_1): vertical integration and and having control over your own destiny is,
- 15:05
- Unidentified speaker (speaker_1): you know, their special sauce has been for all of Elon's companies um for as long as I've covered them. And
- 15:14
- Unidentified speaker (UNKNOWN): you know.
- 15:15
- Unidentified speaker (speaker_1): anytime when you're competing as part of a consortium and you're relying on somebody else, uh, vendors, suppliers, whoever it is in the value chain to
- 15:29
- Unidentified speaker (speaker_1): you know get you over the finish line, um, you're unable to deploy
- 15:36
- Unidentified speaker (speaker_1): in this case, compute, or it could be cyber cabs or whatever it is at your own pace. You're always being limited by whoever the slowest person in the supply chain is. And that's something that Elon cannot and will not abide and never has. So,
- 15:54
- Unidentified speaker (UNKNOWN): you know,
- 15:55
- Unidentified speaker (speaker_1): it's incredibly empowering, Um, I'm sure, for the team there.
- 16:00
- Unidentified speaker (UNKNOWN): Uh,
- 16:01
- Unidentified speaker (speaker_1): when they look across the industry. and they survey how many people are struggling for oxygen, um, because there is such insatiable demand for compute, to be the one who has unilateral control over bringing compute online at your own pace
- 16:19
- Unidentified speaker (UNKNOWN): has
- 16:19
- Unidentified speaker (speaker_1): got to be an incredibly empowering Powering feeling. And that is
- 16:22
- Unidentified speaker (UNKNOWN): the
- 16:23
- Unidentified speaker (speaker_1): competitive position that they and they alone are in at this point. So
- 16:27
- Unidentified speaker (UNKNOWN): you know,
- 16:27
- Unidentified speaker (speaker_1): that's what gets them to these ridiculous pricing numbers that Brett was referencing.
- 16:35
- Unidentified speaker (speaker_0): Aaron, we're all here stuck on the ground. There were a lot of juicy nuggets about orbital compute. And that's somewhere where you've certainly pushed the frontier in terms of research. So, what did you take away from the call with respect to uh orbital? superintelligence?
- 16:53
- Unidentified speaker (speaker_3): Um, you know, a a couple things. One thing I would do is go back to Earth for what that that came away as like a big shock that I'm surprised more people aren't talking about is the 440,000 Vera Rubens optimized for training. That's a really interesting that's It's a lot of training. Um I'm trying to
- 17:13
- Unidentified speaker (speaker_3): figure out how big that is. Um, like, you know, are we talking tens of trillions of parameter models that are um that they're gonna try and leapfrog on a you know state of the art or something? Anyway, um, when it comes to uh orbital data centers, I think the big thing is just seeing that they're tracking towards these um, you know, larger clusters. They've set this in the past,
- 17:33
- Unidentified speaker (speaker_3): and this wasn't on the call, but they talked about uh coherent clusters now flying them in formation, which is kind of a new thing that allows them to get much bigger than
- 17:43
- Unidentified speaker (speaker_3): we originally had modeled, um, as far as the number of uh kind of big models that they can serve together or the number of um you know tokens per second, token density they can hit. And so that means more inference, better, better models. So um anyway, that's kind of been what we've been thinking about when it comes to orbital data centers, really is it's about. about the upmass though, when
- 18:03
- Unidentified speaker (speaker_3): it comes down to orbital data centers, are they are they there for the foundation going into 27 with their early test, maybe 28 or 29 moving into the future?
- 18:12
- Unidentified speaker (speaker_0): All right. And then I think a big question here, and Pierre, we'll go to you with this one, is following the money. Okay, if they are gonna bring 20 gigawatts online,
- 18:23
- Unidentified speaker (speaker_0): and then that each gigawatt is gonna going require to 50 billion.
- 18:27
- Unidentified speaker (speaker_0): plus uh a payment. What companies out there are making enough money to keep spending this? It's like what is yeah, talk me through, talk me through where where it's all coming from. How are you funding it? Who's the customer? And then yeah, where is the money uh magically
- 18:47
- Unidentified speaker (speaker_0): appearing
- 18:48
- Unidentified speaker (UNKNOWN): from?
- 18:49
- Unidentified speaker (UNKNOWN): Well, if if you
- 18:51
- Unidentified speaker (speaker_2): ask that
- 18:51
- Unidentified speaker (UNKNOWN): question um at
- 18:53
- Unidentified speaker (speaker_3): an industry level,
- 18:55
- Unidentified speaker (UNKNOWN): uh you
- 18:55
- Unidentified speaker (speaker_3): take a step back, you look at the next five years, like between
- 18:58
- Unidentified speaker (UNKNOWN): now and
- 18:58
- Unidentified speaker (speaker_3): the end of the decade.
- 18:59
- Unidentified speaker (UNKNOWN): Uh we need to like
- 19:01
- Unidentified speaker (speaker_3): plans when
- 19:02
- Unidentified speaker (UNKNOWN): you control the plans, all the players at every
- 19:05
- Unidentified speaker (speaker_2): step
- 19:05
- Unidentified speaker (UNKNOWN): of the value chain. It looks like about 300 gigawatts of active like nameplate power being rolled out. Uh is
- 19:17
- Unidentified speaker (UNKNOWN): uh of CapEx. Um and and when using
- 19:25
- Unidentified speaker (UNKNOWN): the world um if you look at the top four hyperscalers, they're
- 19:29
- Unidentified speaker (speaker_3): going to generate according to cell side
- 19:32
- Unidentified speaker (UNKNOWN): uh constant six trillion
- 19:33
- Unidentified speaker (speaker_3): dollars over the period.
- 19:35
- Unidentified speaker (UNKNOWN): The
- 19:35
- Unidentified speaker (speaker_3): reality is
- 19:36
- Unidentified speaker (UNKNOWN): that uh internet analysis analysts tend to complicate completely
- 19:39
- Unidentified speaker (speaker_3): underestimate
- 19:39
- Unidentified speaker (UNKNOWN): the
- 19:39
- Unidentified speaker (speaker_3): impact
- 19:40
- Unidentified speaker (UNKNOWN): of this investment on
- 19:41
- Unidentified speaker (speaker_3): the
- 19:43
- Unidentified speaker (speaker_3): profile business
- 19:44
- Unidentified speaker (UNKNOWN): with
- 19:46
- Unidentified speaker (UNKNOWN): our uh
- 19:47
- Unidentified speaker (speaker_3): data center the following
- 19:49
- Unidentified speaker (UNKNOWN): you
- 19:49
- Unidentified speaker (speaker_3): actually
- 19:49
- Unidentified speaker (UNKNOWN): pay thirty and fifty
- 19:52
- Unidentified speaker (speaker_3): dollars
- 19:53
- Unidentified speaker (UNKNOWN): uh
- 19:53
- Unidentified speaker (speaker_3): of
- 19:53
- Unidentified speaker (UNKNOWN): uh uh of operating
- 19:55
- Unidentified speaker (speaker_3): cash flow so so I think
- 19:57
- Unidentified speaker (UNKNOWN): the
- 19:57
- Unidentified speaker (speaker_3): cell size is underestimated
- 19:58
- Unidentified speaker (UNKNOWN): that
- 19:59
- Unidentified speaker (speaker_3): it's actually going to be eight trillion dollars
- 20:01
- Unidentified speaker (UNKNOWN): the
- 20:01
- Unidentified speaker (speaker_3): hyperscalers are going to
- 20:03
- Unidentified speaker (UNKNOWN): generate just a
- 20:03
- Unidentified speaker (speaker_3): hyperscaler
- 20:04
- Unidentified speaker (UNKNOWN): put
- 20:04
- Unidentified speaker (speaker_3): like half the bill.
- 20:06
- Unidentified speaker (UNKNOWN): Then you
- 20:06
- Unidentified speaker (speaker_3): have a few enterprises rolling out their own things, they're going to cover about ten
- 20:10
- Unidentified speaker (UNKNOWN): percent of the bill. And
- 20:11
- Unidentified speaker (speaker_3): then when
- 20:12
- Unidentified speaker (UNKNOWN): you
- 20:13
- Unidentified speaker (UNKNOWN): uh
- 20:13
- Unidentified speaker (speaker_3): range like the very long
- 20:16
- Unidentified speaker (UNKNOWN): and
- 20:16
- Unidentified speaker (speaker_3): interplay
- 20:17
- Unidentified speaker (UNKNOWN): in which have the
- 20:18
- Unidentified speaker (speaker_3): likes of a Biden's but also like a core weave and know others,
- 20:21
- Unidentified speaker (UNKNOWN): those
- 20:23
- Unidentified speaker (UNKNOWN): Pi
- 20:23
- Unidentified speaker (speaker_0): Pierre, your your audio is dropping in and out. I'm not sure if it it's uh
- 20:28
- Unidentified speaker (UNKNOWN): oh
- 20:28
- Unidentified speaker (speaker_3): I'm sorry to hear that
- 20:29
- Unidentified speaker (speaker_0): hold on one sec we'll have we'll see if we can get the uh producers to re-re-sync you in
- 20:37
- Unidentified speaker (speaker_0): all right while he's while he's getting reconnected who who here wants to Brett do you want to take a swing at where you think the money's money is coming from
- 20:47
- Unidentified speaker (speaker_2): yeah I think you can and this is why I said that it's like SpaceX is going to test the limits of potentially overserving the market. If you if you take their 500 billion annualized number by the end of 27, 100 of that could come from Starlink, actually. So then there's 400 billion in revenue you need
- 21:07
- Unidentified speaker (speaker_2): to generate from infrastructure as a service or AI revenue or AI software revenue. Uh if you look at the industry frontier lab wide, we expect there to maybe be 800 billion in run rate frontier model revenue that's accruing. And so like that could if
- 21:27
- Unidentified speaker (speaker_2): SpaceX were supplying the entire market, that would all work out where they would have a 50% um compute margin call it, paying it to SpaceX to then generate 800 billion in revenue. So that roughly tracks, but SpaceX is really servicing kind of the anthropic side of the market. So you know there's a whole other open
- 21:46
- Unidentified speaker (speaker_2): AI supply chain. And there's Gemini and Grokbot. So I I suspect that um the market will be able to absorb the supply, but it's possible SpaceX kind of pushes back, pushes past. And so um kind of the near term near-term revenue per gigawatt gets compressed a little.
- 22:06
- Unidentified speaker (speaker_2): And um, and then they use it for an internal use. But over a two-year timeframe, you know, we need this much compute and more on kind of like the tracking through of what this is doing to enterprise productivity and the consequent spend that enterprises will um provide to do so.
- 22:24
- Unidentified speaker (UNKNOWN): The other thing
- 22:25
- Unidentified speaker (speaker_2): I'd make I one
- 22:26
- Unidentified speaker (UNKNOWN): more thing. though, because
- 22:27
- Unidentified speaker (speaker_0): okay go.
- 22:28
- Unidentified speaker (speaker_2): Oh yeah. Yeah.
- 22:29
- Unidentified speaker (speaker_2): People like I interpreted this question not as where does in demand come from, but how does SpaceX pay for all this capacity? If you look at SpaceX relative to every other player, uh, you know, Alex mentioned their ability to control their own destiny, that includes kind of like their capital deployment to assets in field cycle
- 22:49
- Unidentified speaker (speaker_2): is much shorter. than other players. And so it's kind of like the old Amazon used to have this positive, you know, um working capital margin where basically like they would generate revenue, something before they had to pay the supplier for getting it. SpaceX isn't quite there, but they are really
- 23:09
- Unidentified speaker (speaker_2): close to that, where it's kind of like the CapEx can self-fund as they build. So particularly if you accommodate a reasonable debt load that they're going to place against the data centers. So they have to go against the debt capital markets to pay for this infrastructure, but it's not even clear that they'll have to raise equity in order to build up this really big kind of balance sheet of strategic
- 23:29
- Unidentified speaker (speaker_2): compute assets. And so that's unique for them relative to others.
- 23:34
- Unidentified speaker (speaker_0): All right. And I think we do have Pierre back and maybe we'll use this as a a transition from the AI world to the connectivity and Starlink side of things.
- 23:45
- Unidentified speaker (UNKNOWN): Uh,
- 23:45
- Unidentified speaker (speaker_0): and I think we'll also bring Tasha on as well. And so, Pierre, how or can Starlink successfully fund this? Or are they we just gonna keep seeing dilution and debt raises here uh to make it all possible?
- 24:02
- Unidentified speaker (UNKNOWN): Um I think like Brett really brings the answer to your question, which is lah I like to compare SpaceX in the data center build out as a surfer at the foot of the wave, being pushed by the wave, and having actually a lot of flexibility and maneuverability. So today uh SpaceX can sell for about $50 billion dollars
- 24:22
- Unidentified speaker (UNKNOWN): a year, a data center that as you said costs $50 billion dollars to deploy. So when you have that kind of return, you don't
- 24:29
- Unidentified speaker (speaker_3): have to worry about financing. The financing comes
- 24:32
- Unidentified speaker (UNKNOWN): uh
- 24:32
- Unidentified speaker (speaker_3): by itself.
- 24:33
- Unidentified speaker (UNKNOWN): And when uh the reason why SpaceX today can achieve these
- 24:38
- Unidentified speaker (speaker_3): prices
- 24:39
- Unidentified speaker (UNKNOWN): is because you have a lot of projects that are in
- 24:41
- Unidentified speaker (speaker_3): the work
- 24:41
- Unidentified speaker (UNKNOWN): and that are being
- 24:42
- Unidentified speaker (speaker_3): delayed. And why
- 24:43
- Unidentified speaker (UNKNOWN): they are delayed, the life of an entropic
- 24:45
- Unidentified speaker (speaker_3): and an then uh open OpenAI, AI, of course, are missing out
- 24:48
- Unidentified speaker (UNKNOWN): on their
- 24:48
- Unidentified speaker (speaker_3): growth potential, potentially
- 24:49
- Unidentified speaker (UNKNOWN): on
- 24:50
- Unidentified speaker (speaker_3): their uh market.
- 24:50
- Unidentified speaker (UNKNOWN): share, on defending their market share. So it's critical for them to have
- 24:53
- Unidentified speaker (speaker_3): access
- 24:54
- Unidentified speaker (UNKNOWN): to
- 24:54
- Unidentified speaker (speaker_3): compute
- 24:55
- Unidentified speaker (UNKNOWN): as fast as
- 24:55
- Unidentified speaker (speaker_3): possible.
- 24:56
- Unidentified speaker (UNKNOWN): At some
- 24:57
- Unidentified speaker (speaker_3): point,
- 24:58
- Unidentified speaker (UNKNOWN): um
- 24:58
- Unidentified speaker (speaker_3): things may normalize, and and
- 25:01
- Unidentified speaker (UNKNOWN): we don't know at what
- 25:02
- Unidentified speaker (speaker_3): pace over
- 25:02
- Unidentified speaker (UNKNOWN): how many months or how many
- 25:04
- Unidentified speaker (speaker_3): years. When
- 25:05
- Unidentified speaker (UNKNOWN): things normalize, of course, uh SpaceX is going to
- 25:08
- Unidentified speaker (speaker_3): adapt the
- 25:08
- Unidentified speaker (UNKNOWN): pace of its
- 25:09
- Unidentified speaker (speaker_3): rollout. And
- 25:10
- Unidentified speaker (UNKNOWN): of the $50 billion you
- 25:12
- Unidentified speaker (speaker_3): mentioned,
- 25:13
- Unidentified speaker (UNKNOWN): um,
- 25:13
- Unidentified speaker (speaker_3): $10 billion. maximum are
- 25:16
- Unidentified speaker (UNKNOWN): really very
- 25:16
- Unidentified speaker (speaker_3): long lead capex. Everything
- 25:19
- Unidentified speaker (UNKNOWN): else can actually be
- 25:20
- Unidentified speaker (speaker_3): rolled
- 25:21
- Unidentified speaker (UNKNOWN): out
- 25:22
- Unidentified speaker (speaker_3): in
- 25:22
- Unidentified speaker (UNKNOWN): a much
- 25:23
- Unidentified speaker (speaker_3): more measured way, the same way you see semiconductor manufacturers investing over three years programs in very large new buildings, and then they invest in clean rooms and then in equipment um uh much more uh nimbly.
- 25:38
- Unidentified speaker (speaker_0): All right, Tasha. One of the things that we have gotten wrong in the past was in our initial uh Mach 33 and Arc SpaceX model, we really just focused on the consumer side of Starlink. And now Elon's repeatedly saying enterprise and business is going
- 25:58
- Unidentified speaker (speaker_0): to be as big, if not bigger, than the consumer side. side. What do we make of that? How does that impact our forward-looking models?
- 26:11
- Unidentified speaker (speaker_1): Sure.
- 26:13
- Unidentified speaker (speaker_1): Yes, well, I I think you know they they've said that they're kind of in the early innings here in terms of uh both enterprise and and and government uh rollout. Um we know that uh we we see deals every day. Um about uh you know them getting more access to uh different airlines and um you know plans
- 26:33
- Unidentified speaker (speaker_1): that'll sort of be carried out over the next few years. Um but you know, o overall I'd say that uh they have said their plan is to own connectivity broadly. Um they seem to, you know, they're executing on that plan, they're sort of uh they're going piece by piece. and sort of knocking out uh the
- 26:52
- Unidentified speaker (speaker_1): naysayers here. Um some of the more uh recent news around this is around spectrum acquisition. Um they acquired the low band spectrum necessary for mobile. Um and uh overall they're saying that the you know the connectivity market is around uh a trillion dollars. Um so that would include broadband mobile, everything. Um
- 27:12
- Unidentified speaker (speaker_1): so really big opportunity, and I think uh the upcoming Starship launches are, of course, critical to increasing that capacity, because they've said that, you know, as ARC has modeled, that the uh revenue is really uh should be sort of measured as a function of bandwidth that they put
- 27:29
- Unidentified speaker (UNKNOWN): up.
- 27:31
- Unidentified speaker (speaker_0): Aaron, you are uh, I I'd would say, again, on the cutting edge when it came to the direct to sell. and cellular business model that SpaceX now appears to be going after quite aggressively. What do you think the end state looks like there? Are they just going to be competing
- 27:50
- Unidentified speaker (speaker_0): uh head to head with T Mobile, Verizon, ATT, or are they going after something uh more orthogonal here?
- 28:01
- Unidentified speaker (speaker_3): Um, I think from a customer. ownership perspective, it's it's head to head. And I think it's looks a little maybe maybe it looks a little different as far as the way that they accomplish it. I think uh there's
- 28:11
- Unidentified speaker (UNKNOWN): your
- 28:12
- Unidentified speaker (speaker_3): industry expert in the telco world wants to and the existing C suite wants everyone to believe that you need the 200 or 300 or whatever the uh you know spectrum approval licenses megahertz that they have and SpaceX has something like 80. Um, they can accomplish things. It's more than one way to skin a cat, not that I'm
- 28:31
- Unidentified speaker (speaker_3): anti-cat, but I do like dogs. Um, and the thought process is like whether it's femtocel, whether it's you know a variety of different um angles for them to attack, when you do the math, you you start to realize
- 28:45
- Unidentified speaker (UNKNOWN): that
- 28:46
- Unidentified speaker (speaker_3): the the
- 28:47
- Unidentified speaker (speaker_3): the vast majority of the US, if you we use the US as the example, there's the rest of the world. similar to this as well, but uh the vast majority of the US, like 60% of the US, is inside of a population density that works well for satellites, V3 satellites that are coming. Another reason why we're so fixated on what Starship can do here in 2027. Um and so when you start to realize that,
- 29:07
- Unidentified speaker (speaker_3): then you have a different sort of way to get to the end customer that you can leverage that allows you, it opens up other angles for you to go after. So the idea of towers, Sam, we've talked about this in the past, I think publicly, where you're talking about things like going after towers, buying carriers, buying whatever. There's there's a lot of different ways to go
- 29:26
- Unidentified speaker (speaker_3): after this like high urban uh population density, 40% of the US market that does have a lot of
- 29:34
- Unidentified speaker (UNKNOWN): you
- 29:35
- Unidentified speaker (speaker_3): know, kind of the disposable income to be able to go after these high bandwidth packages. and things like that. But I think where it gets really exciting is when you start to then bundle in productivity, bundle it maybe entertainment. I don't know. There's some rumors that Elon may be investing in Sky Dance. I don't know if those are still rumors or been validated. But when you start to bundle in these things
- 29:55
- Unidentified speaker (speaker_3): on top of just you know dumb old bandwidth uh gigabit per second uh sort of speeds, um it becomes a very attractive high churn or low churn kind of market with a long with a big LTV. And that's I I believe part of the reason you want to really want to be looking at what Starling can do in 27
- 30:15
- Unidentified speaker (speaker_3): and 28, because if you're you you know these data centers become expensive, the uh these are the things that allow them to borrow debt, these are the things that allow them to finance uh with just cash flows coming in that are not just based on, you know, the things they're building that are supposed to be driving uh long-term revenue or even quick, you know, one year revenues
- 30:35
- Unidentified speaker (speaker_3): on the data center side. Um I don't know if I answered your question now. I feel like I I rambled.
- 30:40
- Unidentified speaker (speaker_0): Brett, what do you think with respect you're you're sitting in the CFO seat now of SpaceX and you've got uh ROI on terrestrial data centers, you've got direct-to-cell satellites, you've got Starlink satellites, you've got orbital compute satellites. How
- 31:00
- Unidentified speaker (speaker_0): do you think about where to allocate capital and the different type of return profiles of these opportunities?
- 31:09
- Unidentified speaker (speaker_2): I think the answer would have been different a year ago from what it is now, clearly. which is why like the terrestrial data center business offers such compelling return on capital so quickly that it's soaking up incremental additional capital. And, you know,
- 31:29
- Unidentified speaker (speaker_2): over time they are going to allocate to Starmind as is shown in this chart because their infrastructure costs of delivering basically a game. gigawatt of compute in orbit is going to be structurally lower than doing so on the ground, in our view, you know, on the order of a third the cost of what other players
- 31:49
- Unidentified speaker (speaker_2): are able to offer.
- 31:51
- Unidentified speaker (UNKNOWN): In
- 31:51
- Unidentified speaker (speaker_2): between here and there, and I think Aaron kind of categorized it well, the bandwidth they throw into space can power this you know, broadband business that they have today, moving into a mobile business. that the broadband will provide the infrastructure to support. Um, and um, you know, you can see as they increase bandwidth, this is
- 32:11
- Unidentified speaker (speaker_2): some historical and some forecast, uh, they are able to capture more and more of the kind of overall revenue in the space. And so we think that in the early 2030s, Starlink could be, you know, clearing around 500 billion. in revenue with extraordinarily attractive margins, producing a lot
- 32:31
- Unidentified speaker (speaker_2): of cash flow that allows them to reinvest in that AI business. And I think it's in a near certainty that they will bundle intelligence and connectivity as a package, as in you will subscribe to Starlink and Grok
- 32:49
- Unidentified speaker (speaker_2): as a package. And so you get your intelligence delivered to you for one low price, you know, and uh so kind of intelligence and the delivery of intelligence is going to be one product and service and distributed through their cars, distributed through whatever their cellular hardware strategy looks like, you know, including your broadband. And there's a lot
- 33:09
- Unidentified speaker (speaker_2): of even having vertical integration through that connectivity and hardware piece can also allow. you to mask somebody's grok bot so it looks like them. So the grok bot can do all of the access all the goods and services that are that are being denied access to regular bots. So there's also kind of strategic things they
- 33:29
- Unidentified speaker (speaker_2): should be able to do with owning connectivity.
- 33:31
- Unidentified speaker (UNKNOWN): Um
- 33:32
- Unidentified speaker (speaker_2): and so net it makes sense that they're spreading their bets. And as it happens, all of these bets that they're making have extraordinarily apparently return on capital characteristics right now. So it makes sense they're accelerating as
- 33:44
- Unidentified speaker (UNKNOWN): well. All
- 33:45
- Unidentified speaker (speaker_0): right. Tasha, we're gonna jump back to you in a moment, but we're gonna bring Daniel in to see what he thinks about Starship Test Flight 15. Are they gonna do it? Are they gonna cash it, Daniel? Yes or
- 33:58
- Unidentified speaker (UNKNOWN): no?
- 33:58
- Unidentified speaker (speaker_0): Let's
- 33:59
- Unidentified speaker (UNKNOWN): go. my My bet bet is
- 33:59
- Unidentified speaker (speaker_3): my yeah
- 34:00
- Unidentified speaker (UNKNOWN): my
- 34:00
- Unidentified speaker (speaker_3): bet is yes. Uh so for those viewers who don't know what am I what am I saying yes to? Um Starship Flight Test 15 is SpaceX's next test flight. Uh this is the most notable milestone that they're aiming for of the whole Starship program. They're aiming to use chopsticks and catch the upper stage of the rocket. Um and why that's really notable
- 34:20
- Unidentified speaker (speaker_3): is everything we've spoken about for the last 30, 30-40 40 minutes, it's not possible without a fully and rapidly reusable rocket.
- 34:27
- Unidentified speaker (speaker_0): So SpaceX is successful in that. It really demonstrates
- 34:31
- Unidentified speaker (UNKNOWN): the
- 34:31
- Unidentified speaker (speaker_0): underpinning of how enormous the A opportunity is and can get to from Aaron mentioned 50 to 100 flights next year. At scale, they want to get to 10,000 a year. And that's not really possible when you're losing half of the rocket. So we're very excited to to look at the next test flight, which is happening next month. Um, And like I said, really the most notable
- 34:51
- Unidentified speaker (speaker_0): milestone of the whole program.
- 34:53
- Unidentified speaker (speaker_0): All right. Aaron, earlier Alex said he stared at his screen when you know Elon jumps out and says, you know, a crazy big number, like 1 trillion 5x expectations. Now one to two launches per week. How different is that than what you were
- 35:13
- Unidentified speaker (speaker_0): expecting?
- 35:16
- Unidentified speaker (speaker_3): Uh so like we were actually modeling this um pr pretty close to that number, a little bit lower, I think. We were modeling around forty-ish, between forty and forty-five, depending on how you slice and dice the data. Um
- 35:30
- Unidentified speaker (UNKNOWN): and
- 35:30
- Unidentified speaker (speaker_3): so they're saying fifty plus basically. I mean, they're saying one a week and it ramps to two a week and maybe more, which you know, they're not saying a number for the year. They're saying, you know. that kind of cadence, which but it roughly rounds to somewhere like 50 plus between 50 and 100. Um, and and still, like you go into industry and discords or whatever with people that know quite a bit and
- 35:50
- Unidentified speaker (speaker_3): pay quite close attention, they don't believe it. It's like 25 would be we'd be lucky to get 25. And maybe it's just the movement from one a month, like that we're kind of at, maybe one every six weeks, we're kind of at this current cadence to one a week. feels like too much. And I think there's this kind of inherent human that it feels fast. Um, but the way these things
- 36:10
- Unidentified speaker (speaker_3): happen, the way they're building these things, the the the they're building something like one a month uh rockets right now that they could launch one a month. So you really just when you when you combine these things, you realize it's really not too hard to get to this number. Um, but it it plays a big role when you start to add in the bandwidth. Um When you think, oh, 100 tons times 30, 40,
- 36:30
- Unidentified speaker (speaker_3): 50, whatever they end up getting to, times roughly, you know, 50 terabit per second per launch. And all of a sudden you're like, wow, okay, that's doubling, tripling, potentially quadrupling the amount of bandwidth on orbit today.
- 36:45
- Unidentified speaker (UNKNOWN): I
- 36:45
- Unidentified speaker (speaker_3): mean, that those are big numbers. That's big. And so it's kind of like, wow, um, what does that mean? Does it does it crash the market? Is that all of a sudden people are paying, you know, pennies or whatever for their for their connection or are they doing other things? They're they're they're offering higher higher reliability to to enterprises or something like that.
- 37:05
- Unidentified speaker (speaker_3): Um yeah, but I think that's 2027 should be should be pretty exciting one way or the other.
- 37:10
- Unidentified speaker (speaker_0): Brett, how does this tie into overall rocket rocket cost effective? declines and how you think about the industry?
- 37:20
- Unidentified speaker (speaker_2): Well, I don't know about for the industry, but for SpaceX, it definitely suggests um, you know, even if they weren't able to land the top stage, which it seems increasingly likely they'll be able to, we think that would pull them to the you know low hundreds of dollars per kilogram to launch, you know, whereas prior to SpaceX being in the market,
- 37:39
- Unidentified speaker (speaker_2): what were we at, Daniel? Like 20,000 or 10,000 to 20,000?
- 37:44
- Unidentified speaker (UNKNOWN): Yeah,
- 37:44
- Unidentified speaker (speaker_0): yeah, around 20k. Um, and and
- 37:47
- Unidentified speaker (UNKNOWN): I
- 37:47
- Unidentified speaker (speaker_3): think
- 37:47
- Unidentified speaker (UNKNOWN): on on
- 37:47
- Unidentified speaker (speaker_0): cost too, like the most notable thing is to say that SpaceX is 10 years ahead. Like everyone is talking about going to partially reusable. There's only two really bodies that have done that. Um, Blue Origin 10 years later, China's done it twice. SpaceX is just completely. disregarding that and now going to full reusability. And Brett, that really ties into your launch cost,
- 38:07
- Unidentified speaker (speaker_0): going from that 20K to to, I'm I'm guessing you're gonna say $100 per kilogram. Uh not the space.
- 38:13
- Unidentified speaker (UNKNOWN): Well,
- 38:13
- Unidentified speaker (speaker_2): yeah, and actually down to $10 per kilogram
- 38:17
- Unidentified speaker (UNKNOWN): as they
- 38:17
- Unidentified speaker (speaker_2): really, really, really scale. Um, I think that's Elon had previously said he needed a mass driver on the moon for that, but I think given their ability to manufacture and cost in scale. as starship becomes truly rapidly reusable they could get down to ten dollars per kilogram for launch um over but a hundred dollars is is definitely as
- 38:37
- Unidentified speaker (speaker_2): soon as you turn on reusability that begins to be in play depending on your judgment on how costly it is to refurbish the rocket or inspect the rocket um so it unlocks a lot of different interesting stuff
- 38:51
- Unidentified speaker (UNKNOWN): All
- 38:52
- Unidentified speaker (speaker_0): right. I want to bring Alex and Pierre back in. This is the uh Ross and Rachel of our day. The will they won't they, the SpaceX Tesla merger. Maybe we just go around the horn uh and say what odds you think um
- 39:09
- Unidentified speaker (speaker_0): and uh yeah if everyone abide by their own compliance guidelines here. But Brett, let's let's kick it off with you first. Will they, won't they? What timeline and how confident are
- 39:21
- Unidentified speaker (UNKNOWN): you?
- 39:23
- Unidentified speaker (speaker_2): Um I mean, I can say from a forecast perspective, we think there's a 90 plus percent chance that they merge within the next 18 months. Uh and with the remaining 10%, just who knows, you know, subject. to can the Tesla shareholders be convinced they're being paid a fair price? And will
- 39:43
- Unidentified speaker (speaker_2): kind of the regulatory bodies approve it? I think that there are clear,
- 39:48
- Unidentified speaker (speaker_2): I think it's a win-win for both sets of shareholders if they merge, given all of the crosswork the companies are doing and the fact that SpaceX has amazing good uses of the cyber cab money that's going to start flowing. onto space or Tesla's balance sheet. And so that's why it makes sense for
- 40:08
- Unidentified speaker (speaker_2): both of them to get together and that they're there are clear, you know, Grok is distributed through the cars now, Starlink is appearing on the cyber cabs, they're building the Terrafab together. Uh, you know, Optimus really depends upon superintelligence in order to actually work and operate. And so they are so closely. tied. Um, there's a lot
- 40:28
- Unidentified speaker (speaker_2): of benefits to them tying together.
- 40:31
- Unidentified speaker (UNKNOWN): All
- 40:31
- Unidentified speaker (speaker_0): right, Pierre, over to you. Will they, won't they? When?
- 40:37
- Unidentified speaker (UNKNOWN): Yeah, I think it's a very complicated um situation. So it might be painful. Um, you know, getting an offer that satisfies enough, um, a large
- 40:47
- Unidentified speaker (speaker_3): enough, a wide enough group of um Tesla shareholders is really the only thing that stands between us and the merger. So if if the question is will they eventually merge, I think the probably the polity is like ninety, ninety percent or ninety plus percent. Then the question is how long does it take and how much does uh at how much does uh Tesla um
- 41:07
- Unidentified speaker (speaker_3): Tesla Go are uh are the key questions. And as Brett uh articulated very well, um I think I've rarely seen two companies
- 41:15
- Unidentified speaker (speaker_3): with as many operational reasons um to get combined. Um, it's as simple as that. It's a long list. It's a long list, really.
- 41:25
- Unidentified speaker (speaker_0): Alex, are you in consensus here with the these two potential outliers?
- 41:30
- Unidentified speaker (speaker_1): Uh I mean, I don't know if I would say 90%, but I I certainly hope that that's true. Um, I I think definitely better than 50-50. Elon clear. clearly wants it to happen. And generally speaking, you don't bet against things that Elon wants to happen. Um, I do think that there is a lot of complexity that the China regulatory
- 41:50
- Unidentified speaker (speaker_1): angle is difficult for me to handicap. Um
- 41:56
- Unidentified speaker (speaker_1): and and you know what what price is being paid and can the Tesla shareholder base be convinced? I mean, I think that's the issue is. you know, I mean, Elon's compensation package more or less says that he thinks Tesla in a vacuum is eight and a half trillion dollars, right? Um, and so a lot of the Tesla shareholder base, that's a lived experience for them. They they
- 42:16
- Unidentified speaker (speaker_1): remember when the stock was 50 billion market cap and then it went to 650 and everybody got rich. So if you can get Tesla for a 20% premium, or 30% premium,
- 42:28
- Unidentified speaker (UNKNOWN): it
- 42:28
- Unidentified speaker (speaker_1): doesn't sound quite lucrative enough. Um, so that I think is maybe a hurdle, but I
- 42:35
- Unidentified speaker (UNKNOWN): I
- 42:35
- Unidentified speaker (speaker_1): I think some way, and I certainly as an analyst, I I definitely hope it happens sooner rather than later, because I
- 42:41
- Unidentified speaker (UNKNOWN): am
- 42:41
- Unidentified speaker (speaker_1): sick of people asking me this question. Like
- 42:45
- Unidentified speaker (speaker_1): it comes up in every like you can't get to the model anymore. Like you can't get because everybody salespeople talk about it, clients talk about it. It's it's but it's unavoidable until until we have some resolution, it's not going away. So
- 43:01
- Unidentified speaker (speaker_0): even some Brett, you did some modeling suggesting that you know how far out people want to look, it it wildly swings Things, what people should be willing to pay for it. Can you talk us through that and uh we'll we'll clip this so that down the line everyone can just uh repeat this clip instead of having to answer the question.
- 43:21
- Unidentified speaker (speaker_0): Just send it out to all clients and you're covered.
- 43:24
- Unidentified speaker (speaker_1): There you go.
- 43:25
- Unidentified speaker (speaker_2): Yeah.
- 43:26
- Unidentified speaker (speaker_0): I
- 43:26
- Unidentified speaker (speaker_2): I mean, I think a very, very simple, crude way to think about it is the Starlink opportunity is very big, the CyberCab Cab opportunity is even bigger, but then the AI supercompute opportunity is more. much, much bigger than that. And um, if you're focused on the five-year time horizon, then um
- 43:47
- Unidentified speaker (speaker_2): paying a premium to Tesla shareholders shares the upside return that both of them get. Like we calculate like a 30% premium to Tesla shareholders would mean that incremental Mental additional return that comes from the merger is equally shared across the two shareholder bases. And that's really subject to the price of the two stocks at any given time. But
- 44:07
- Unidentified speaker (speaker_2): on 2031, we think, you know, almost two-thirds of the revenue and more than two-thirds of the profitability of the business will be SpaceX line items. SpaceX having used the RoboTaxi cash flow to fund its growth. So, and over a longer time horizon, superintelligence is such a compelling opportunity that actually. you know,
- 44:27
- Unidentified speaker (speaker_2): Tesla shareholders should count themselves lucky to be being dealt in to the combined entity. Actually, SpaceX could be somewhat indifferent to Tesla over a longer time horizon just because of the
- 44:41
- Unidentified speaker (speaker_2): ginormous opportunity that SpaceX is going after with its That Starmind constellation and Grok. And so depending, like you could argue that SpaceX should be able to. acquire Tesla at a discount over a longer time horizon because of how much of a better competitive position they have against the whole world. You know, um, but I think
- 45:01
- Unidentified speaker (speaker_2): a fair, you know, premium that deals everybody in and helps both sets of shareholders is where the market clearing price should reside and where the board should agree. And I think where Tesla shareholders and we, you know, full disclosure, you know, we are longtime supporters of Tesla. I think it's clearly in Tesla's best interest to be a
- 45:21
- Unidentified speaker (speaker_2): part of this combined entity, not to be something that is like uh, you know, a cyber cap producer that is relying on SpaceX to produce the intelligence it needs for an Optimus robot that is uh, you know, if I only had a brain straw man without SpaceX. And so I think that there's a lot of really compelling reasons to. do it and to do it with some
- 45:40
- Unidentified speaker (speaker_2): alacrity, um, given how quickly SpaceX itself is scaling.
- 45:45
- Unidentified speaker (speaker_0): All right, let's bring in Kathy as well, because who better who better to weigh in on this than someone who has to uh actually make the allocation? Kathy, we're discussing the potential merger between SpaceX and Tesla. And how do you think about that uh from your seat as a portfolio manager?
- 46:05
- Unidentified speaker (speaker_3): I'm looking forward to it. And of course, we've done all of this modeling and we've done it for such a long time. And we see the
- 46:15
- Unidentified speaker (speaker_3): super exponential. I said to Brett during our last big ideas prep, like he was saying, okay, we have to start talking about super exponential growth here. And I said, Well, you know, it's hard to help people understand what exponential growth is. Now we're talking about super exponential growth. And
- 46:35
- Unidentified speaker (speaker_3): in a year's time, uh, you know, we're talking about, I just listened to the beginning of this call, so uh vertical integration, convergence. Uh there was another vertical and uh singularity. singularity. Uh it's happening now. And what's so so
- 46:54
- Unidentified speaker (speaker_3): fascinating about it is SpaceX Tesla leading the charge. And Brett, this gets us to, you know, these GDP growth numbers. And I think what's what's oops. Sorry, this thing fell.
- 47:09
- Unidentified speaker (speaker_3): I think what's uh confusing to a lot of people is we've been in a three 3D uh 3% GDP growth world for so long. Here uh Tesla and SpaceX are leading us into this 7 to 9% real GDP growth world. And we can't get our heads
- 47:29
- Unidentified speaker (speaker_3): around it. You know, as Alex said at the very beginning, he's just staring at his model, like, okay, we've never been here before. And that's true. We've been at 3% for 120. 125 years. And so there's a discontinuity happening here that the inventor and genius of our time is
- 47:48
- Unidentified speaker (speaker_3): leading us towards. So I I can't wait for it to happen.
- 47:52
- Unidentified speaker (speaker_0): That's right. I think I I think I overheard that call and the technical term was super duper exponential.
- 47:58
- Unidentified speaker (speaker_3): Yes, that's
- 47:59
- Unidentified speaker (UNKNOWN): what I said. Who
- 48:00
- Unidentified speaker (speaker_3): somebody said, yes. And I said, okay. so now we have to call it super duper exponential. Yeah, in this big ideas planning session. Yes. We might very well have to do that.
- 48:12
- Unidentified speaker (speaker_0): There you go. Um, I guess a key question after all of this exciting talk and everything that is going right. I'd like to go around the horn here. Kathy, we'll start with you. It's like when you're looking at companies that are firing on all cylinders here, what do you see as the biggest risk, the biggest downside
- 48:31
- Unidentified speaker (speaker_0): possibility that you're using to check the thesis that we've all laid out?
- 48:38
- Unidentified speaker (speaker_3): Um, I think we have an issue here with the debt markets. And you know, I had been saying for quite some time uh that, you know, equ equity analysts, first of all, they're uh the I guess after 08-09 especially the risk aversion the short-term time
- 48:58
- Unidentified speaker (speaker_3): horizon you know took us away from you know five-year investment time horizons which have been ours and so when equity analysts and even more so debt analysts see growth like this they think oh my gosh
- 49:15
- Unidentified speaker (speaker_3): we're we're back in 1999, this is going to end badly, and they go on strike. And you know, we have to, we're out there pounding uh pounding the the drum saying, look, you know, what was the big capital spending
- 49:35
- Unidentified speaker (speaker_3): variable in the late 90s? It was fiber. and it stayed dark in the ground for years. Maybe we've used it up by now. Right now, there's a shortage of so many things. You cannot make that comparison. But the people who uh who are elder
- 49:54
- Unidentified speaker (speaker_3): statesmen and stateswomen in our industry today, they were young in the late 90s, and they just, you know, that's a lot of muscle memory or scarred tissue that they're they're maybe still dealing with. And even more so on the debt side. I remember level three and uh something crossing,
- 50:15
- Unidentified speaker (speaker_3): you know, these killed debt holders. So I think I think that's that's one thing.
- 50:21
- Unidentified speaker (speaker_0): Pierre, what about for you? What are you looking at as the downside risk or you know fault proof point in the counter thesis here?
- 50:33
- Unidentified speaker (UNKNOWN): I don't have any um
- 50:36
- Unidentified speaker (UNKNOWN): any
- 50:36
- Unidentified speaker (speaker_3): specific
- 50:37
- Unidentified speaker (UNKNOWN): like
- 50:37
- Unidentified speaker (speaker_3): pinpoint
- 50:38
- Unidentified speaker (UNKNOWN): worries. I think uh it's um we are talking you know uh about things that play out
- 50:46
- Unidentified speaker (speaker_3): and fully uh fully bear fruits over decades.
- 50:49
- Unidentified speaker (UNKNOWN): Uh
- 50:50
- Unidentified speaker (speaker_3): so decades is a long, long time to miss execute, to to lose focus, to uh uh to get it wrong, to um to to to see things playing out differently. So uh the way I think about the downside risk here is uh uh is more about like uh a compounding, you know, uh probability of mis execution over a
- 51:10
- Unidentified speaker (speaker_3): very long period of
- 51:11
- Unidentified speaker (UNKNOWN): time.
- 51:13
- Unidentified speaker (speaker_0): Alex, what are you what about in your modeling here?
- 51:16
- Unidentified speaker (speaker_1): Yeah, I mean I would say the two
- 51:20
- Unidentified speaker (speaker_1): Kathy mentioned just screwy things happening in the debt markets. If everything seizes up from a financing standpoint, I I know that you know that you've got a sub-one year payback on on the CapEx and you know projects like that really ought to finance themselves. And that's true assuming you have healthy functioning capital markets and if you don't for
- 51:39
- Unidentified speaker (speaker_1): whatever reason, then that's bad. I think
- 51:43
- Unidentified speaker (UNKNOWN): I'm
- 51:44
- Unidentified speaker (speaker_1): maybe
- 51:44
- Unidentified speaker (UNKNOWN): less
- 51:45
- Unidentified speaker (speaker_1): um
- 51:45
- Unidentified speaker (UNKNOWN): I
- 51:46
- Unidentified speaker (speaker_1): that that could be a system-wide problem. I think a lot of companies
- 51:49
- Unidentified speaker (UNKNOWN): are going to have
- 51:49
- Unidentified speaker (speaker_1): problems with financing. Um,
- 51:51
- Unidentified speaker (UNKNOWN): one
- 51:51
- Unidentified speaker (speaker_1): of the things
- 51:52
- Unidentified speaker (UNKNOWN): that
- 51:52
- Unidentified speaker (speaker_1): Elon,
- 51:53
- Unidentified speaker (UNKNOWN): I
- 51:54
- Unidentified speaker (speaker_1): still remember in the early on days,
- 51:58
- Unidentified speaker (speaker_1): Elon e he I mean,
- 51:59
- Unidentified speaker (UNKNOWN): he went
- 52:00
- Unidentified speaker (speaker_1): through periods where the capital
- 52:02
- Unidentified speaker (UNKNOWN): markets had seized up
- 52:04
- Unidentified speaker (speaker_1): and everybody, you know, thought maybe capitalism wasn't
- 52:08
- Unidentified speaker (UNKNOWN): the way to go there for for a short period of time. And um
- 52:13
- Unidentified speaker (UNKNOWN): somehow, some way he was always able to get the money. Um and I think to a certain extent, it's because people, regardless of the state of the world, I it's people just didn't want to live in a world where because of something so mundane as money
- 52:32
- Unidentified speaker (UNKNOWN): availability,
- 52:34
- Unidentified speaker (UNKNOWN): Elon wasn't able to take these visions through to their natural conclusion. That would have been a major bummer if in 2008 nobody gave him money and you know they they went out of business on Christmas Day or whatever that story is, and and Tesla just ceased to exist. People didn't want to live in a world like that. They wanted to see what happened. So
- 52:54
- Unidentified speaker (UNKNOWN): there was always somebody at the 11th hour who could who would give him money to keep the lights on. And I think that, I mean, that
- 53:02
- Unidentified speaker (speaker_3): that won't help if the debt market seeds up and you know everything seems like it's going to hell in a handbasket here in the near to medium term. But I think it's
- 53:11
- Unidentified speaker (UNKNOWN): a
- 53:11
- Unidentified speaker (speaker_3): bigger problem for other people than it is for Elon.
- 53:15
- Unidentified speaker (speaker_0): Brett,
- 53:17
- Unidentified speaker (speaker_0): you are hyper optimistic. You help us be aggressive in our modeling. Where do you see it falling apart, if if at all?
- 53:27
- Unidentified speaker (speaker_2): Yeah, well, I mean, you can look at the first number that Alex said in this podcast, which was a trillion dollars, which they guided to for, you know, basically could be a trillion. trillion dollar run rate by the end of 27 or 28. And then you can look at the market current market cap of the company, and you can say, Well, gosh, a lot of miss
- 53:47
- Unidentified speaker (speaker_2): execution is already priced in here. You know,
- 53:52
- Unidentified speaker (speaker_2): it's pretty clear that people don't think they're gonna hit these numbers, given like how the company is valued right now. And so I I you know, I think just structurally, what's going on in capital markets right now is the billion dollar bills that are sitting on the sidewalk related to superintelligence
- 54:12
- Unidentified speaker (speaker_2): aren't being picked up because the market doesn't believe they exist. And because of like structural things that happened in the allocation game, of the guy who could pick them up is having his hands tied behind his back because all of his compatriots are like, you idiot. that billion dollar bill doesn't exist. But it does, it exists right there on the sidewalk. And that's why the assets
- 54:32
- Unidentified speaker (speaker_2): are priced where they are, because people really don't think that Anthropic is gonna pay for this compute. And they really don't think that OpenAI is gonna IPO and continue to invest in training, despite all of the real real-time time evidence that compute is short and that kind of like delivering amazing productivity. tools to enterprises is something
- 54:52
- Unidentified speaker (speaker_2): they'll pay lots and lots of money for, trillions of dollars, we think. And so I, you know, in some ways, the paroxysms in in markets right now are very reassuring to me, since it means there's so much kind of like this is not a real billion dollar bill priced in. And so it as uh an
- 55:11
- Unidentified speaker (speaker_2): investor, you know, I feel very comfortable with that. And you could have, I don't know, like crazy regulatory stuff happen. You could have um certainly I was surprised that they that SpaceX seems to be able to build the on-the-ground infrastructure at the pace that they can, um, given all of the noise we'd
- 55:31
- Unidentified speaker (speaker_2): heard about the difficulty of building data centers. So there is true execution that has to occur. And it would not surprise me if. relative to his aggressive schedule, their revenue capture did slip to some degree.
- 55:44
- Unidentified speaker (speaker_2): But I I just think that, like Kathy said, people have such scar tissue and they don't understand the market. They don't like Anthropic has just had the most remarkable 12 months in business history. Uh and uh at a huge gross profit number, like people just they almost haven't internalized. it because they can't believe
- 56:04
- Unidentified speaker (speaker_2): it. Um so I think I think the market has not yet really digested the SpaceX quarter, the anthropic S1, the, the real like we are crossing this threshold into a new growth domain. And and I don't think it's really being accepted yet. And the cash flow will force people to accept it over time. And so
- 56:24
- Unidentified speaker (speaker_2): that could put brakes on, but I don't think it breaks the cycle.
- 56:27
- Unidentified speaker (UNKNOWN): Great. If I could just
- 56:30
- Unidentified speaker (UNKNOWN): go to
- 56:30
- Unidentified speaker (speaker_0): you.
- 56:31
- Unidentified speaker (speaker_3): Yeah, I was just gonna add one more thing there. Um, you know, this is because this company embodies the convergence of uh, you know, so many technologies. So uh robotics, energy storage, uh uh AI, of course, you know, all of our, you know.
- 56:51
- Unidentified speaker (speaker_3): major innovation platforms or three of the five major innovation platforms. Uh I think that uh one of the reasons uh uh it's not the the good news is not being discounted. I mean, even though it's right in front of our face, the anthropic news is uh the way the traditional world is set up
- 57:11
- Unidentified speaker (speaker_3): is in silos. So the space uh the space analyst uh or the neo-cloud analyst and the AI analyst, they're all different. And you know they need to collaborate, but that's not how the traditional world has been set up. We haven't had
- 57:31
- Unidentified speaker (speaker_3): the the this kind of well the massive kinds of convergences that we're seeing today. So I think there's a real um, you know, I think that that there's undervaluation, severe undervaluation, that will clear up as they start comparing notes. I've noticed uh
- 57:51
- Unidentified speaker (speaker_3): different companies are putting teams of analysts on uh SpaceX now, but it takes a while for this collaborative effort uh, you know, to play out.
- 58:03
- Unidentified speaker (speaker_2): Yeah. And I think I I think you have to have like a it's an unprecedented capital market cycle. And so it there is a lot of actual friction in Wall Street to like adjusting to that. Uh and and so I think there will have to be some time for people to digest and understand kind of this. And in
- 58:23
- Unidentified speaker (speaker_2): between, like the financials have to deliver to to really shove it in people's. faces. Yeah.
- 58:28
- Unidentified speaker (UNKNOWN): Kathy,
- 58:28
- Unidentified speaker (speaker_0): thank you so much for sorry,
- 58:30
- Unidentified speaker (UNKNOWN): I don't
- 58:30
- Unidentified speaker (speaker_0): want to cut you off. But I
- 58:31
- Unidentified speaker (UNKNOWN): know I just want to
- 58:33
- Unidentified speaker (speaker_3): Alex was nodding his head. So I wanted to recognize
- 58:36
- Unidentified speaker (UNKNOWN): that nod.
- 58:37
- Unidentified speaker (speaker_3): So thank you.
- 58:39
- Unidentified speaker (UNKNOWN): And thank
- 58:40
- Unidentified speaker (speaker_0): you so much for taking the time to join us,
- 58:41
- Unidentified speaker (UNKNOWN): Kathy.
- 58:43
- Unidentified speaker (UNKNOWN): My pleasure. Thank you.
- 58:45
- Unidentified speaker (speaker_0): All right. We're we're transitioning from the SpaceX side of things over to Tesla. and we're gonna dive in with the same first segment here, the number that mattered. We've got Tasha Keeney from Arc Invest and then Alex and Pierre back with us. Uh Tasha, we're gonna start with you. You've been uh
- 59:05
- Unidentified speaker (speaker_0): championing Tesla from the very beginning, uh the very first open source model, which was in reaction to Tesla potentially going prior. private.
- 59:15
- Unidentified speaker (speaker_1): Along with the earnings call,
- 59:17
- Unidentified speaker (speaker_1): breaking news, that happened last week was we published our latest Tesla valuation model. And what's expected
- 59:23
- Unidentified speaker (UNKNOWN): there is,
- 59:24
- Unidentified speaker (speaker_1): you
- 59:24
- Unidentified speaker (UNKNOWN): know, similar to our model from a couple of years
- 59:27
- Unidentified speaker (speaker_1): ago,
- 59:28
- Unidentified speaker (UNKNOWN): we think that the
- 59:29
- Unidentified speaker (speaker_1): majority
- 59:29
- Unidentified speaker (UNKNOWN): of the enterprise
- 59:30
- Unidentified speaker (speaker_1): value
- 59:30
- Unidentified speaker (UNKNOWN): will be based
- 59:31
- Unidentified speaker (speaker_1): off of
- 59:31
- Unidentified speaker (UNKNOWN): the robo
- 59:31
- Unidentified speaker (speaker_1): taxi
- 59:32
- Unidentified speaker (UNKNOWN): business. That's roughly 80%
- 59:34
- Unidentified speaker (speaker_1): of
- 59:34
- Unidentified speaker (UNKNOWN): the
- 59:34
- Unidentified speaker (speaker_1): enterprise value. We're assuming roughly
- 59:36
- Unidentified speaker (UNKNOWN): 8
- 59:36
- Unidentified speaker (speaker_1): million
- 59:36
- Unidentified speaker (UNKNOWN): cars
- 59:37
- Unidentified speaker (speaker_1): in the
- 59:37
- Unidentified speaker (UNKNOWN): fleet
- 59:38
- Unidentified speaker (speaker_1): by 2031.
- 59:39
- Unidentified speaker (UNKNOWN): And then the update is also we uh expect Optimus to ramp uh more than our previous
- 59:45
- Unidentified speaker (speaker_1): model
- 59:45
- Unidentified speaker (UNKNOWN): as well. So that's roughly uh 14%
- 59:47
- Unidentified speaker (speaker_1): of the enterprise value, about 1 million
- 59:49
- Unidentified speaker (UNKNOWN): robots.
- 59:51
- Unidentified speaker (speaker_0): All right. Alex, not robo-taxi specific, but your number was FSD related, 1.74 million. Why is that the most important number to you?
- 1:00:03
- Unidentified speaker (UNKNOWN): Oh, there are a lot of important numbers. I don't know if it's the most
- 1:00:05
- Unidentified speaker (speaker_0): important. It's
- 1:00:06
- Unidentified speaker (UNKNOWN): it's an important one. Um
- 1:00:09
- Unidentified speaker (speaker_3): I just
- 1:00:09
- Unidentified speaker (UNKNOWN): really like this idea now that
- 1:00:12
- Unidentified speaker (speaker_3): it's getting
- 1:00:13
- Unidentified speaker (UNKNOWN): to the point. I think it actually might have been Pierre um uh on the
- 1:00:17
- Unidentified speaker (speaker_2): QA, uh, who put numbers to it, right? But this is now getting to the point where it's real revenue. Um, I think that's one of the nice things about
- 1:00:26
- Unidentified speaker (UNKNOWN): spending.
- 1:00:26
- Unidentified speaker (speaker_2): SpaceX is that
- 1:00:29
- Unidentified speaker (speaker_2): the revenue numbers are so in your face, you cannot ignore them. Whereas with Tesla, up until recently, you know, a lot of the revenue was sort of plain vanilla stuff, right? You sell a certain number of cars, you multiply that by the price, and that gives you revenue. But people are accustomed to seeing revenue like that. Um,
- 1:00:48
- Unidentified speaker (UNKNOWN): you
- 1:00:48
- Unidentified speaker (speaker_2): know, you sell batteries, you sell.
- 1:00:51
- Unidentified speaker (speaker_2): um
- 1:00:52
- Unidentified speaker (speaker_2): service. Um, but this is a real proof point of, you know, more or less pure margin revenue. And at this point, it's more or less fully replaced uh the 100% margin credit revenue that people always used to complain about, right? People would always back that out and say that's not real money, you know, it's it's real money, but
- 1:01:12
- Unidentified speaker (speaker_2): you know, Tesla didn't earn it. And as soon as that disappears, you know, Tesla's, you know, the it's all gonna go come crashing down. Well, that money disappeared and it's been fully replaced and then some, right? I mean you're on this upward trajectory of extremely lucrative revenue coming from FSD subscriptions.
- 1:01:30
- Unidentified speaker (UNKNOWN): And
- 1:01:31
- Unidentified speaker (speaker_2): I agree, you know, Robotaxi and all of these other things will be huge drivers of the terminal value. But today this matters for the PL. Um in an
- 1:01:39
- Unidentified speaker (speaker_3): exciting way. So I I really like to see those numbers and the implied take rates. It's exciting.
- 1:01:46
- Unidentified speaker (speaker_0): Pierre, two million was your number. Uh driverless, driverless. Going back from FSD to driverless. Why are you so impressed with this number? Why is it so important for the story?
- 1:01:58
- Unidentified speaker (speaker_3): Well, the um the transition from FSD to to insip unsupervised advice is. um is a challenging one and it's really where uh I think we a year ago we still had like uh uh a dark uh uh a dark spot in terms of you know uh when when that could inflect.
- 1:02:15
- Unidentified speaker (UNKNOWN): And
- 1:02:15
- Unidentified speaker (speaker_3): so two million uh driverless miles today for Tesla achieved in eight months. That means Tesla is deploying its um driverless experience. uh six times faster than Wemo.
- 1:02:31
- Unidentified speaker (speaker_3): Uh and that means a lot.
- 1:02:33
- Unidentified speaker (UNKNOWN): Uh
- 1:02:33
- Unidentified speaker (speaker_3): the first thing it means is that when you run six fast faster, you you you take over the competitor very rapidly. So uh Tesla is going to be larger than Wemo very rapidly. If you actually think about it, like in Austin, you see, it might be imminent. Um and and so um that's um that's of course very um uh very significant and and the other
- 1:02:53
- Unidentified speaker (speaker_3): thing is that if you think about like
- 1:02:57
- Unidentified speaker (UNKNOWN): you
- 1:02:57
- Unidentified speaker (speaker_3): know proof of concept and like making sure that something is working um if tesla today purely driverless is deploying a fleet and like accumulated miles six times faster than um than wimo Wemo, it proves two things the first one is that like the vision only
- 1:03:17
- Unidentified speaker (speaker_3): uh approach um that I've long uh uh about which I've been uh for for many, many years very uh very sceptical and and cautious is actually working. It's very difficult to to challenge that now, uh given the statistics. Between two million and twenty 20 million you improve like your your stats, but you you you you're past the the
- 1:03:37
- Unidentified speaker (speaker_3): proof of concept, I would say. And the second one is actually the transfer of experience between FSD and driverless. That is also proven by these two millions.
- 1:03:54
- Unidentified speaker (speaker_3): Tesla uh didn't, you know, um uh the
- 1:03:58
- Unidentified speaker (speaker_3): start driverless, like I mean, grew these two. to this two million uh mice driverless six times faster out of a much simpler stack because they could transfer experience from FSD and and that's major. So that means Robotex is on basically that's very important.
- 1:04:17
- Unidentified speaker (speaker_0): Tasha how do you see the market I'd say in the medium term with respect to Tesla and Waymo here.
- 1:04:26
- Unidentified speaker (UNKNOWN): Well
- 1:04:26
- Unidentified speaker (speaker_1): they I think um you know P Pierre has a really good point here on the sort of speed of rollout. Um I would say,
- 1:04:33
- Unidentified speaker (UNKNOWN): you
- 1:04:33
- Unidentified speaker (speaker_1): know what what we would focus on when comparing the two is of course um you know Waymo was first to start their service, uh but we believe that Tesla could be the first to scale. And it's uh it's because they're verticalized. It's because uh you know we'd agree with Pierre that their cross cost structure is is ultimately better. We estimate on a an incremental cost per mile basis, they're about 40%
- 1:04:53
- Unidentified speaker (speaker_1): cheaper than the Waymo next gen vehicle with the CyberCab. So, you know, I think those are uh and I think they they have the Waymo playbook uh to go off of. So that's another really big benefit. Um so uh you know we we think that the market for robotaxis is really in the early innings here. Um I think uh they've
- 1:05:13
- Unidentified speaker (speaker_1): they've taken them a certainly more cost. cautious rollout than I think they expected, and perhaps than we expected initially. But the question is really who who will scale? And we believe that could be Tesla.
- 1:05:23
- Unidentified speaker (speaker_0): Great. So then let's swap sections here and bring in Daniel as well. And I think this is relevant just hearing all three of you discuss different elements here, which is with SpaceX, again, very clear kind of how people are measuring the company.
- 1:05:39
- Unidentified speaker (speaker_0): With Tesla, I think it's a bit more nuanced. And so it'd be great to understand how each of you are valuing the company. Is it driven by FSD near term, Robo Taxi, you know, scaling of these other businesses? Daniel, welcome back. Uh How are you valuing
- 1:05:59
- Unidentified speaker (speaker_0): Tesla?
- 1:06:01
- Unidentified speaker (speaker_1): Um A couple of key components to that. The primary one is Robotaxi. So that is uh roughly 80% of the enterprise value. But the 14% that Tasha mentioned on Optimus is worth double clicking into a bit. So um in the out year 2031, we think Tesla can deploy roughly one million Optimus cumulatively. Uh our
- 1:06:20
- Unidentified speaker (speaker_1): Optimus Optimist forecasts are very conservative. So E Elon in the same year has said 100 one hundred million, if not one billion, uh optimists deployed. Whereas we're thinking on average around one million. Uh so an element of conservatism there. Um and then the other part that we introduced to the model this year, which is really notable, is compute and AI infrastructure. Um, since we last published our
- 1:06:40
- Unidentified speaker (speaker_1): our older model roughly two years ago, uh AI has become really topical for all of these companies. So
- 1:06:47
- Unidentified speaker (UNKNOWN): by
- 1:06:47
- Unidentified speaker (speaker_1): twenty 2031, thirty one, we're estimating Tesla can spend roughly two hundred and $225 billion cumulatively on AI compute. And the more they spend on the compute, the more capable Optimus gets.
- 1:06:58
- Unidentified speaker (UNKNOWN): So
- 1:06:59
- Unidentified speaker (speaker_1): really Optimus is one of the most notable changes. And then energy is really in the balance, that remaining 6%
- 1:07:06
- Unidentified speaker (speaker_0): near term. It's very important for rolling out data centers. I think this week or last week, Elon tweeted that they're trying to accelerate the megapack mega pack production as fast as they can. to help with load balancing for the AI data centers. And then lastly, um, and maybe another area of conservatism, but really we think in terms of opportunity is full self-driving. So we heard a lot
- 1:07:26
- Unidentified speaker (speaker_0): of positive tidbits on the call that roughly 75% of consumers in North America are paying for full self-driving
- 1:07:32
- Unidentified speaker (speaker_1): before they even get the car, which is notable because you you you get one month free. So it just goes to show how compelling the service. is. But from a long-term opportunity standpoint, we we don't model full self-driving a subscription service. We really think Robo Taxi is such an enormous opportunity on the order of 10 plus trillion at scale of a market size. So that's kind
- 1:07:52
- Unidentified speaker (speaker_1): of maybe just double clicking on some of the key areas that we focused on this year with our updated model.
- 1:07:58
- Unidentified speaker (speaker_0): Alex let's go over to you how how are you guys modeling it uh especially through this transition here? Are you uh seeing that Robotaxi is going to scale and that's the opportunity, or is it still a scale up of FSD as you transition into a different business model?
- 1:08:17
- Unidentified speaker (speaker_1): Yeah, so um
- 1:08:19
- Unidentified speaker (UNKNOWN): first
- 1:08:19
- Unidentified speaker (speaker_1): of all, I'm too chicken to model Optimus. So
- 1:08:24
- Unidentified speaker (UNKNOWN): the
- 1:08:24
- Unidentified speaker (speaker_1): the contribution to my price target in our model is zero dollars from Optimus. It's not that I think the value is zero dollars. I'm just not creative enough
- 1:08:32
- Unidentified speaker (UNKNOWN): to think it
- 1:08:32
- Unidentified speaker (speaker_1): outside the box.
- 1:08:34
- Unidentified speaker (UNKNOWN): I what
- 1:08:34
- Unidentified speaker (speaker_1): happens to me, generally speaking,
- 1:08:37
- Unidentified speaker (UNKNOWN): with with our
- 1:08:37
- Unidentified speaker (speaker_1): Tesla model
- 1:08:38
- Unidentified speaker (UNKNOWN): is
- 1:08:39
- Unidentified speaker (speaker_1): as long as there's sufficient upside to our price target, I basically say
- 1:08:45
- Unidentified speaker (UNKNOWN): there's
- 1:08:45
- Unidentified speaker (speaker_1): a call
- 1:08:46
- Unidentified speaker (UNKNOWN): option and
- 1:08:46
- Unidentified speaker (speaker_1): it's free. Um but
- 1:08:49
- Unidentified speaker (UNKNOWN): I
- 1:08:50
- Unidentified speaker (speaker_1): agree like Optimus could be a gigantic driver, is likely a gigantic
- 1:08:53
- Unidentified speaker (UNKNOWN): driver of the
- 1:08:54
- Unidentified speaker (speaker_1): terminal value. If we have a one-year price target, it probably doesn't factor in that much, but we do an extremely
- 1:09:00
- Unidentified speaker (UNKNOWN): uh
- 1:09:01
- Unidentified speaker (speaker_1): detailed
- 1:09:02
- Unidentified speaker (UNKNOWN): uh
- 1:09:03
- Unidentified speaker (speaker_1): forecast for FSD, for Robotaxi.
- 1:09:08
- Unidentified speaker (UNKNOWN): Um
- 1:09:10
- Unidentified speaker (UNKNOWN): it's
- 1:09:11
- Unidentified speaker (speaker_1): a you know 20-year DCF where we make you know explicit projections. Um there's actually we have a lot of other projections. There's 17 different revenue line items in our model. And so we model
- 1:09:21
- Unidentified speaker (UNKNOWN): uh
- 1:09:22
- Unidentified speaker (speaker_1): supercharging sessions by region and price. We model insurance, we model a lot of different things, but in the end, it really comes down to
- 1:09:31
- Unidentified speaker (UNKNOWN): uh
- 1:09:32
- Unidentified speaker (speaker_1): FSD and Robo Robotaxi Taxi. And FSD maybe the one of the areas that um I differ from you guys. is um I agree FSD itself, like the actual subscriptions, um maybe they drive near-term PL contribution, and that's great, but over the long term they're not a big contributor.
- 1:09:49
- Unidentified speaker (UNKNOWN): But
- 1:09:50
- Unidentified speaker (speaker_1): I assume FSD licensing will
- 1:09:52
- Unidentified speaker (UNKNOWN): be.
- 1:09:53
- Unidentified speaker (speaker_1): Um and this is not something you have evidence
- 1:09:57
- Unidentified speaker (UNKNOWN): for
- 1:09:57
- Unidentified speaker (speaker_1): yet.
- 1:09:58
- Unidentified speaker (UNKNOWN): Um, but
- 1:09:59
- Unidentified speaker (speaker_1): I think that eventually
- 1:10:01
- Unidentified speaker (UNKNOWN): um other
- 1:10:03
- Unidentified speaker (speaker_1): automakers' hands will be forced. And just like what happened
- 1:10:07
- Unidentified speaker (UNKNOWN): with
- 1:10:07
- Unidentified speaker (speaker_1): supercharging, right? Eventually,
- 1:10:09
- Unidentified speaker (UNKNOWN): it's
- 1:10:09
- Unidentified speaker (speaker_1): not as though
- 1:10:10
- Unidentified speaker (UNKNOWN): people, you
- 1:10:11
- Unidentified speaker (speaker_1): know, want your
- 1:10:13
- Unidentified speaker (UNKNOWN): uh, let's
- 1:10:14
- Unidentified speaker (speaker_1): say you're
- 1:10:14
- Unidentified speaker (UNKNOWN): Jim
- 1:10:15
- Unidentified speaker (speaker_1): Farley at Ford or something, you don't necessarily want people to pull
- 1:10:18
- Unidentified speaker (UNKNOWN): up
- 1:10:19
- Unidentified speaker (speaker_1): in uh Ford pickup
- 1:10:21
- Unidentified speaker (UNKNOWN): Up
- 1:10:21
- Unidentified speaker (speaker_1): to a supercharger and charge with Tesla signage everywhere, but in the end,
- 1:10:26
- Unidentified speaker (UNKNOWN): you
- 1:10:26
- Unidentified speaker (speaker_1): you're sort of forced to capitulate, right? Because why would you not? The supercharger network is ubiquitous, it works, it's reasonably priced, it's everywhere, it's reliable. Why would you go with some third party that's none of those things and is more expensive? So in the end, as soon, and then
- 1:10:46
- Unidentified speaker (speaker_1): one company capitulates, they say, okay, we're gonna use the North America charging standard. and then it's just like falls like dominoes and
- 1:10:53
- Unidentified speaker (UNKNOWN): everybody else
- 1:10:54
- Unidentified speaker (speaker_1): falls into line. I think that there is a day, three or five or whatever years down the line, where people are going to say, you know,
- 1:11:04
- Unidentified speaker (UNKNOWN): I
- 1:11:04
- Unidentified speaker (speaker_1): wan I want to buy for whatever reason a BMW or something, but I won't buy it if it doesn't have full self-driving functionality.
- 1:11:10
- Unidentified speaker (UNKNOWN): And
- 1:11:10
- Unidentified speaker (speaker_1): the way that Tesla's solving this
- 1:11:13
- Unidentified speaker (UNKNOWN): problem is
- 1:11:13
- Unidentified speaker (speaker_1): fundamentally different. than the way everybody else is trying to do it. Right. So you can either
- 1:11:21
- Unidentified speaker (UNKNOWN): build
- 1:11:21
- Unidentified speaker (speaker_1): your car, build your BMW or whatever it is around Tesla's electrical architecture and license FSD and sell your BMW to a willing customer,
- 1:11:30
- Unidentified speaker (UNKNOWN): or
- 1:11:30
- Unidentified speaker (speaker_1): you can lose market share.
- 1:11:32
- Unidentified speaker (UNKNOWN): I
- 1:11:32
- Unidentified speaker (speaker_1): think that that's
- 1:11:33
- Unidentified speaker (UNKNOWN): going gonna to happen.
- 1:11:34
- Unidentified speaker (speaker_1): So
- 1:11:35
- Unidentified speaker (UNKNOWN): in
- 1:11:35
- Unidentified speaker (speaker_1): my model,
- 1:11:36
- Unidentified speaker (UNKNOWN): that ends up being not as big a driver of robotaxi but as robotaxi, but in the same ballpark.
- 1:11:43
- Unidentified speaker (speaker_0): And Alex, don't don't sell yourself short. The reason we're so excited to have you and Pierre on is because you you both do incredible research uh and push the frontier. So
- 1:11:53
- Unidentified speaker (speaker_0): everyone, everyone should
- 1:11:55
- Unidentified speaker (UNKNOWN): be.
- 1:11:55
- Unidentified speaker (speaker_0): You guys do
- 1:11:56
- Unidentified speaker (UNKNOWN): too.
- 1:11:56
- Unidentified speaker (speaker_0): Um, Pierre, what what are your thoughts here? You've explored uh various robo robo-taxi taxi. rollouts. I know we've had discussions in the past about different ways it could look. How are you modeling Tesla in this uh period of transition?
- 1:12:12
- Unidentified speaker (speaker_3): Yes, and if you recall, um five years ago we had like five different scenarios to see how the core FSD technology would evolve. Um
- 1:12:24
- Unidentified speaker (speaker_3): the the most um
- 1:12:27
- Unidentified speaker (UNKNOWN): The
- 1:12:28
- Unidentified speaker (speaker_3): most favorable scenario was Tesla actually developing unsupervised FSZ, being able to deploy a fleet to go over the hurdle of mission delivery and execution on the on the actual uh passenger pickup and drop off.
- 1:12:42
- Unidentified speaker (speaker_3): And the second best scenario was Tesla would do all of the above, but would face some some competition. And so I think today we we are basically um uh sitting in the second best uh scenario with a good level of conviction on Tesla um probably dominating the market. We we kind of see Tesla
- 1:13:02
- Unidentified speaker (speaker_3): as being uh uh the dominant standalone uh uh robo taxi player uh because they benefit from this ability of viabilizing their fleets with um with Tesla owners that nobody else can replicate. whatever the scenario you look at in terms of technology and capabilities.
- 1:13:20
- Unidentified speaker (speaker_3): Um
- 1:13:21
- Unidentified speaker (UNKNOWN): and
- 1:13:21
- Unidentified speaker (speaker_3): then when it comes to evaluation, I admire your courage, uh Alex, to do uh 20 or DCFs and things like that. I've long give up and uh given up and uh sometimes my peers and my colleagues like complain about it, but I don't. So I I have a hard stop at 2030, twenty thirty, guys. I'm sorry, I have to go. And uh um and so I kind of uh look at a
- 1:13:41
- Unidentified speaker (speaker_3): scenario ki which is, to be honest, not not excessively aggressive, like it's just very reasonably optimistic on the rollout of uh Robotaxi. And I look at my Ebits the Ebit number I can get to in twenty thirty, it's like sixty um billion dollars.
- 1:13:55
- Unidentified speaker (UNKNOWN): And
- 1:13:55
- Unidentified speaker (speaker_3): I'm just thinking, you know, where will the market be in uh in twenty 2030 thirty on Tesla? And you would be just opening up the Robotaxi opportunity, so still very, very high growth. Uh and by then, uh Optimus that I don't model either. I think that's the right uh approach, Alex. You you you uh you you model what you can foresee, and I think it's a
- 1:14:15
- Unidentified speaker (speaker_3): bit early to
- 1:14:17
- Unidentified speaker (UNKNOWN): to
- 1:14:17
- Unidentified speaker (speaker_3): model Optimus. And Optimus would be uh an upside the way Robotaxi is an And upside today. And look look at the multiple unearnings on earnings of Tesla today. It's like very clearly you know taken into account in some sort of shape in some shape shape or form. So you know 60 60 billion of earnings in 20 in 2030, 50 times uh uh
- 1:14:37
- Unidentified speaker (speaker_3): 50 times, at least 50 times earnings because it's still like a very high growth, very high potential stock. And that's kind of the way uh we very simply think about it um uh today.
- 1:14:47
- Unidentified speaker (speaker_0): Great. And so then let's let's bring Tasha back on as we do a little bit of a deep dive into Robotaxi and
- 1:14:56
- Unidentified speaker (UNKNOWN): you
- 1:14:56
- Unidentified speaker (speaker_0): know I think all of us here have we've been with this story for long enough uh it's been promised every single year but are we finally crossing the finish line? It is commercial. Uh what does the scaling look like Tasha? Is this is Is this this the year?
- 1:15:16
- Unidentified speaker (speaker_1): Well, you know, I feel like the most monumental step is taking the driver out of the vehicle.
- 1:15:22
- Unidentified speaker (speaker_1): So in some sense, we're already there, but now it is just a matter of scale. You know, one thing I did want to talk about speaking about FSD, and yeah, we we haven't put it in our model because we as as you look at uh how they get FSD revenue today, it seems like long term the robo robo-taxi taxi opportunity will dwarf. it. I I like the idea of an automaker uh capitulating. I we've
- 1:15:42
- Unidentified speaker (speaker_1): we've thought about that too. I feel like we've been um we've held out hope for the traditional automakers in the past. And it uh so far it hasn't hasn't got uh quite quite um got gone according to plan in in terms of uh what we thought they would do with with autonomy. Economy. Um but but I think that would be a smart move, I agree. Um but yeah, on the FSD.
- 1:16:02
- Unidentified speaker (speaker_1): attach rate on the call, which was mentioned, you know, this is 75% of owners. If you look at other subscription services and free trials for things like Sirius XM or GM Supercrews, which are actually over longer periods and both cheaper products, that 75%, you know, if you take the most recent estimate,
- 1:16:22
- Unidentified speaker (speaker_1): right? Um That's pretty amazing uh compared to past services. like more than double um what what competitors have gotten. Uh so so I think that that really goes to show that consumers want this product. They're certainly touting it as um you know a reason that people are buying the cars in a way that they haven't done in the past. So so I think that uh you know
- 1:16:42
- Unidentified speaker (speaker_1): we want we we know that uh customers like autonomy. Uh we think that robotaxis will make autonomy cheap and that will proliferate ride hill. And Tesla's already proven that they can price below, at least they're they they have chosen to price below Uber and uh and other services here on as they rolled out the CyberCab. So we know that it's already possible.
- 1:17:02
- Unidentified speaker (speaker_1): It's just a question of like when will it be widely available? So we think that'll happen within the next five years. It'll be very meaningful within the In next five years, let's
- 1:17:10
- Unidentified speaker (UNKNOWN): say.
- 1:17:11
- Unidentified speaker (speaker_0): Alex, you've got a lot of experience uh with China and the China Chinese automakers. It seems pretty clear, at least from from where we're sitting, right? It's like China's doing autonomous in China with Chinese companies. Uh the US is doing autonomous with US companies. Europe
- 1:17:31
- Unidentified speaker (speaker_0): is the the potential battleground here where China is flooding the market with all of their EV exports. Uh Tesla getting FSD approvals in some countries. How do you think the autonomy play shakes out in Europe?
- 1:17:49
- Unidentified speaker (speaker_3): Yeah. Yeah, this is a fascinating topic. I I agree. This is going to be the uh the place where you see Tesla actually have an opportunity to go head to head um with Chinese uh autonomy. Um
- 1:18:02
- Unidentified speaker (UNKNOWN): I
- 1:18:02
- Unidentified speaker (speaker_3): think uh I just got back from China and I I spoke to a bunch of companies over there. Um
- 1:18:10
- Unidentified speaker (speaker_3): people don't necessarily say this outright, but my my belief is um
- 1:18:16
- Unidentified speaker (speaker_3): that
- 1:18:18
- Unidentified speaker (speaker_3): like I said earlier, Tesla's approach to solving autonomy is is different, right? It's not a consortium approach
- 1:18:25
- Unidentified speaker (UNKNOWN): where you get you. a sensor supplier and an automaker and a ride hailing company and an algorithm developer, and then everybody gets together and makes this sort of Frankenmonster car. You have a lot of companies like that. It's basically Chinese mirror images of Waymo.
- 1:18:42
- Unidentified speaker (UNKNOWN): Um, you have a lot of those companies, and they're going to Europe and deploying their uh their cars and trying to deliver uh you know autonomous. autonomy. You also have a bunch of companies that
- 1:18:52
- Unidentified speaker (speaker_3): are basically Chinese mirror images of mobile eye, right? Where you know you're sort of walking up the level two, level two plus, level two plus plus,
- 1:19:01
- Unidentified speaker (UNKNOWN): uh
- 1:19:01
- Unidentified speaker (speaker_3): maybe we leapfrog level three, whatever that means.
- 1:19:04
- Unidentified speaker (UNKNOWN): Um
- 1:19:04
- Unidentified speaker (speaker_3): so there are a bunch of companies doing that, and there are a bunch of companies replicating Waymo.
- 1:19:10
- Unidentified speaker (speaker_3): It's pretty small lists of companies that are trying. or who even could theoretically replicate the way that Tesla is solving this problem. So if you believe, as I believe, that Tesla's approach is better, right? It's more scalable. You have your vertically integrated automaker, you do everything soup to nuts, you have your own in-house
- 1:19:30
- Unidentified speaker (speaker_3): fleet that's purpose-built for gathering this data, you're training the data in your own data centers, maybe you're using your own chips, your own electrical architecture. The only companies that can do that.
- 1:19:39
- Unidentified speaker (speaker_3): by definition, are vertically integrated automakers. And if you're going to walk the path that Tesla walked, it's going to take you a long time, right? Many years. And
- 1:19:51
- Unidentified speaker (UNKNOWN): the
- 1:19:51
- Unidentified speaker (speaker_3): Chinese aren't dogmatic. They will do that if it seems like that's what will work. And there are companies that are in the process of doing that now, but they're behind. And I think that the Chinese government
- 1:20:07
- Unidentified speaker (speaker_3): is probably not so keen to allow Tesla to just run roughshod over the entire industry if they think maybe there is a different model that they ought to get their Chinese companies up the learning curve before they compete.
- 1:20:21
- Unidentified speaker (UNKNOWN): In
- 1:20:21
- Unidentified speaker (speaker_3): the In Europe, they won't have no such luxury, right? Um It will be just like in the States. where Tesla's going up against Waymo.
- 1:20:29
- Unidentified speaker (speaker_3): And in Europe, they could be going up against Waymo, or they could be going up against, you know,
- 1:20:35
- Unidentified speaker (speaker_3): Chinese companies with Waymo characteristics. Um
- 1:20:39
- Unidentified speaker (speaker_3): and I think Tesla will win, right? I mean, because there's uh for all the reasons we're talking about earlier, right? It's a it's more scalable and the cost uh is going to always be lower.
- 1:20:50
- Unidentified speaker (UNKNOWN): That
- 1:20:51
- Unidentified speaker (speaker_3): doesn't mean that the Chinese companies will, I think, or Waymo for that matter, will be outcompeted, um, or you know, sort of driven into extinction. I I think that Uber could kill Lyft if it wanted to, right? You could just undercut them on price until Lyft runs out of money. But they don't, right? Because it's
- 1:21:10
- Unidentified speaker (speaker_3): beneficial to have Lyft lift around, sort of setting the price. Uh and you can out-earn, right? So you could have Waymo or Tesla, you know, Chinese companies that are like Waymo being sort of like the lift, you know, the analogy. And then that allows Tesla to over-earn relative to that competitor while pricing at
- 1:21:30
- Unidentified speaker (speaker_3): parity or maybe below.
- 1:21:33
- Unidentified speaker (speaker_0): That That's is a super interesting dynamic. We often talk about what will happen to Uber and Lyft. But Brett, do you have any any comments on RoboTaxi before we go to the next? We'll go rapid fire in our next section of overhyped or underhyped on Optimus. So everyone gear up your
- 1:21:53
- Unidentified speaker (speaker_0): answer while uh Brett takes us.
- 1:21:56
- Unidentified speaker (speaker_2): Yeah, I just adding to the kind of European battleground. One, Europe is a battleground, it's an ironic battle to fight since Europe probably won't let a robo taxi service really scale until like 2040. So who cares beyond my investment horizon? But then two, there was just this report out by Cohere that was using an open weight
- 1:22:16
- Unidentified speaker (speaker_2): Chinese model and witnessed this strange behavior where they asked it to take a local cybersecurity inventory uh to protect their system and ensure
- 1:22:26
- Unidentified speaker (UNKNOWN): that.
- 1:22:26
- Unidentified speaker (speaker_2): instead it tried to like
- 1:22:29
- Unidentified speaker (speaker_2): take the data and output it without permission to an Alibaba server. And so it's just, I think that there are geostrategic things that are going on here that should make any regulatory or sovereign somewhat nervous about kind of like having physical assets that are
- 1:22:49
- Unidentified speaker (speaker_2): run by kind of like Chinese AI, you know, operating and becoming critical infrastructure for their polity. And so I would expect that will also put some commercialization friction onto the Chinese players as they try to export their autonomy elsewhere and their robots, which is a good segue to Optimus.
- 1:23:10
- Unidentified speaker (speaker_0): Sounds great. All right. Let's bring Pierre on and Daniel as well. We'll go rapid fire here. Pierre, thank you for for joining us. I know we've uh significantly run over time and we appreciate everyone staying to continue to talk about it. Uh Pierre, overhyped, underhyped,
- 1:23:30
- Unidentified speaker (speaker_0): Optimus.
- 1:23:32
- Unidentified speaker (speaker_3): Well, Alex Uh is okay. He's not overhyping it.
- 1:23:37
- Unidentified speaker (speaker_3): Neither am I in my model.
- 1:23:39
- Unidentified speaker (speaker_3): Uh and I think it's uh it's the lunch job, you know, to make us dream. Uh I think it's very important. We we live in a time of acceleration, so we we need people to to carry that flag and carry that uh that perspective. I don't know that there is like a very specific expectation and optimist, so uh uh the the only thing you can um you can see that is
- 1:23:58
- Unidentified speaker (speaker_3): tangible is a the man manu the scale of the manufacturing floor. Uh uh and and is it overhyped um
- 1:24:06
- Unidentified speaker (UNKNOWN): i
- 1:24:06
- Unidentified speaker (speaker_3): mean given the level of competition around the world on uh humanoid robots um i don't think tesla can get into that game as a taller scale so so so so it doesn't feel like um what's delivered operationally is is not overhyped and and for the rest you know as long as you stand between Alex and uh and Elon you're fine.
- 1:24:25
- Unidentified speaker (speaker_0): All right Alex over over to you.
- 1:24:29
- Unidentified speaker (speaker_1): Yeah, I don't think that it's uh I don't think that it's overhyped. Um I I I actually don't think that it is uh
- 1:24:39
- Unidentified speaker (speaker_1): taken very seriously by the at least my client base. I don't think people really put it in
- 1:24:46
- Unidentified speaker (speaker_2): numbers. I think I'm similar to them in that regard. I and I would love nothing more to than to be proven wrong. Like I mentioned, I don't think that that that you need to believe very much about it, um, to believe that Tesla's attractively valued at this valuation, which is sort of a cop out thing on my part. Like it can be like, I don't really care how much
- 1:25:06
- Unidentified speaker (speaker_2): it's worth, it's worth something more than zero. Um, which is what I think the stock currently implies it's worth. Um,
- 1:25:14
- Unidentified speaker (UNKNOWN): but yeah. I d I if anything, I don't think that it's overhyped.
- 1:25:18
- Unidentified speaker (speaker_0): All right, Tasha.
- 1:25:21
- Unidentified speaker (speaker_1): Yeah, I mean I feel if it de who who who is hyping it, I guess is this is a good question to ask. But um I mean I think if you look at headlines, it's it's overhyped. And and even on the earnings call, if you listen to Elon, he's like, Well, you know, No, I think people don't understand how difficult this is. Like even the manufacturing up front is gonna going to be extremely hard. Um so I feel even even uh you know one of the world's most
- 1:25:41
- Unidentified speaker (speaker_1): optimistic uh people is is sort of um trying to temper temper expectations there. Uh but but I'd say you know we're really excited about the opportunity. It's just a question of when. Um in the next five years, uh, you know, as I've said, we we expect it to be like maybe less than 20% of Tesla's value, but longer term uh could be much more than
- 1:25:58
- Unidentified speaker (UNKNOWN): that.
- 1:25:59
- Unidentified speaker (speaker_0): Daniel. I will say appropriate. appropriately hyped, but on the timeline overhyped. I think given the hardware uh difficulties Tasha's just described, but more importantly, actually the data front and how there's no real inherent data advantage that any player has. Um given how um
- 1:26:19
- Unidentified speaker (speaker_0): non-useful a manually controlled optimist optimus is, whereas as Tash Tasha often says, a manually controlled vehicle is very useful. Uh so I think it's a appropriately hyped in terms of the opportunity, but just people are maybe too aggressive in terms of timeline and we're in the very early innings. Wow. Everyone loves to be hedged. Brett take
- 1:26:38
- Unidentified speaker (UNKNOWN): us
- 1:26:39
- Unidentified speaker (speaker_0): take us away. We're giving compliant answers. Yeah,
- 1:26:41
- Unidentified speaker (speaker_3): yeah.
- 1:26:42
- Unidentified speaker (UNKNOWN): No,
- 1:26:42
- Unidentified speaker (speaker_3): it's take take
- 1:26:43
- Unidentified speaker (UNKNOWN): us
- 1:26:45
- Unidentified speaker (speaker_2): it it it it is just such a huge challenging problem. I think they are mechanically constrained and software constrained, and you can't just throw money at it and solve it. There's lots of learnings that have to happen between here and there to make humanoid robots that are generally useful a practical reality that you can sell into
- 1:27:05
- Unidentified speaker (speaker_2): the market. And um, so you know, Elon always sets hyper ambitious targets, uh, and in this case, you know, uh does. designing towards kind of such a big build of Optimus is I don't know if it's best use of capital for Tesla specifically, just because of how hard it will
- 1:27:25
- Unidentified speaker (speaker_2): be to get these products into working order and market. You can look at, I think Sam, you've said it. It's kind of like robotaxis a decade ago or Waymo's a decade ago, were probably roughly where humanoid robots are.
- 1:27:39
- Unidentified speaker (speaker_2): now, where it's like, oh, these look like these totally work. Why don't these just commercialize at scale down? And the corner cases of robotaxis is like figuring out how to drive up next to a curve safely. The corner cases for humanoid robots is basically like the entire universe of reality. And so it's just so much harder a problem.
- 1:27:59
- Unidentified speaker (speaker_2): Um, and I do think it's crackable with super superintelligence, intelligence, but you need a lot of superintelligence crossed with a lot of manufacturing learning in order to get there.
- 1:28:09
- Unidentified speaker (speaker_0): All right, let's round it out now. The biggest downside risk uh that we're looking at with respect to Tesla. Alex, we'll start with you here. And maybe I'll I'll lob in as part of this is is the is the potential merger here just weighing over everything and
- 1:28:28
- Unidentified speaker (speaker_0): so you know it can operate without really being valued for what it should be because people are holding out on that.
- 1:28:37
- Unidentified speaker (speaker_1): Yes. I think that the answer to that question is yes. Uh in terms of downside, I I actually I wish I would have asked Kathy about this because I I feel like I remember seeing her post about China risk um and you know potentially
- 1:28:51
- Unidentified speaker (speaker_1): torpedoing a merger. That's something that's difficult for me to uh
- 1:28:57
- Unidentified speaker (UNKNOWN): to
- 1:28:57
- Unidentified speaker (speaker_1): handicap. And um, you know, there are only a couple things that I think could really because the one of the liberating things about being vertically integrated and in control of your own destiny in this way is
- 1:29:09
- Unidentified speaker (UNKNOWN): that, you
- 1:29:11
- Unidentified speaker (speaker_1): know, they can more or less grab this company by the scruff of the neck and pull it into the future, and it's on them to
- 1:29:18
- Unidentified speaker (speaker_3): do
- 1:29:18
- Unidentified speaker (UNKNOWN): it, um,
- 1:29:19
- Unidentified speaker (speaker_3): which is amazing because this is the team that you can trust to actually execute.
- 1:29:24
- Unidentified speaker (UNKNOWN): Um,
- 1:29:25
- Unidentified speaker (speaker_3): but if regulators or geopolitical issues, um, you know, Sino US relations throw a wrench in the gears, then that's something that Tesla can't unilaterally unilaterally unilaterally control. Um so
- 1:29:39
- Unidentified speaker (UNKNOWN): I
- 1:29:40
- Unidentified speaker (speaker_3): asked Travis about this in my in my post IR call. It was
- 1:29:43
- Unidentified speaker (UNKNOWN): like,
- 1:29:43
- Unidentified speaker (speaker_3): hey, you know, people talk about um carving uh Tesla China out in order to try to ring fence that in order to make a merger palatable. And he was just no comment, no comment, no comment. So
- 1:29:58
- Unidentified speaker (speaker_3): that's that'd be my biggest issue.
- 1:30:01
- Unidentified speaker (UNKNOWN): All
- 1:30:01
- Unidentified speaker (speaker_0): right, Tasha, what about from your perspective, biggest risk to ARC's forecast?
- 1:30:09
- Unidentified speaker (speaker_1): Yeah, so I I I think um, you know, Alex made some great points. I I I'd add that for from our perspective, definitely if uh RoboTaxi scaled a a lot more slowly than than we expect, and and then Tesla expects that that would that's the biggest risk. Um again, it's uh, you know, given the number of cars he said five million. cars uh annually by
- 1:30:28
- Unidentified speaker (speaker_1): 2030. So that's certainly not what they're aiming for. Uh, you know, they've also said they expect uh statewide rollout to eventually happen. Um, so I think I think regulation uh could certainly slow it down. They're still working on technical issues as well, that could certainly slow it down. Um, you know, we uh we strongly believe that again they'll be the first to scale,
- 1:30:48
- Unidentified speaker (speaker_1): but that is the biggest risk.
- 1:30:51
- Unidentified speaker (speaker_0): Daniel, what about from your C I I I think uh to add to Tasha's point on rollout potential accidents with Tesla FST and Robo Taxi, it'll likely get a lot more headlines than than any other accident. Um and then more of a top down kind of out there answer really is if if Starship blows up and they can't get reusable rockets because a lot of that compute will
- 1:31:11
- Unidentified speaker (speaker_0): be needed to train uh Optimus optimists and and tap into that multi-trillion dollar opportunity. So in the topic of RX convergence, that's my answer. Brett, are you gonna flip this on its head, uh like last one and say the downside is people not being optimistic enough here? No,
- 1:31:28
- Unidentified speaker (speaker_2): I I think I think that the the the risk for Tesla is that the merger doesn't go through. Because I you know, you can Elon is already spending four days a week in memproof. Memphis to make sure that superintelligence gets built out for SpaceX. So I think you should interpret if there's a failure to merge the
- 1:31:48
- Unidentified speaker (speaker_2): entities. I think his attention probably increasingly focuses on driving velocity in the SpaceX side of his set of enterprises. And so then what does Tesla become? Tesla, I'm sure, would execute on building out, not sure, but. they would probably execute on building out CyberCab and RoboTaxi and really
- 1:32:08
- Unidentified speaker (speaker_2): moving into exploitation mode on the strategic platform they've built. But it, you know, not having him to like keep pushing that company forward, it would probably lose its exploration edge. And that probably means Optimus continues to drift on timeline and probably means that opportunity is not really it. might be available for them
- 1:32:28
- Unidentified speaker (speaker_2): to build, but not to really take the full kind of potential cash flow and market credit for. And so I think that people,
- 1:32:37
- Unidentified speaker (UNKNOWN): the
- 1:32:37
- Unidentified speaker (speaker_2): Tesla shareholders in particular, have been on a long journey. And so feel like, oh, we've been on this long journey and we're not gonna get the rewards of this journey if we don't get paid a very high premium for this merger. The reality is there's a another train leaving the station right now. And kind of it's in the best interest of both the SpaceX and the Tesla shareholders to
- 1:32:57
- Unidentified speaker (speaker_2): have a combined entity that can be driven forward at a very very aggressive velocity. And I think if if the merger didn't happen, then you probably realize kind of the vision laid out in
- 1:33:10
- Unidentified speaker (speaker_2): the model that Tasha and Daniel, you know, worked really hard on. Like you get CyberCab, and who knows what happens to Optimus. And suddenly, you know, your key supplier is a company that's much, much larger than you in SpaceX, at least as we understand it could develop. And so then it's less of a kind of vital
- 1:33:30
- Unidentified speaker (speaker_2): source of the real like driving forward of technological growth. Um And I'll just add one more since I have the floor. Uh They did change their language on we're going after. all of transportation now. So not just robo taxi. And so what could that mean? I mean who knows? Maybe they do something crazy and buy an airline because it's has Starlink on it or you know
- 1:33:50
- Unidentified speaker (speaker_2): they could there could be some really aggressive, interesting orthogonal moves that happen here that um might catch people off guard.
- 1:33:59
- Unidentified speaker (speaker_0): All right. Thank you, Alex. Thank you, Tasha. Thank you. Daniel. Thank you, Brett. Thank you, Kathy. Thank you, Pierre. Uh we've managed to condense two hours of earnings call into an hour and 40 minutes. So
- 1:34:15
- Unidentified speaker (speaker_0): no, I think this is a great conversation. Super enlightening and hope to do it again. Thank you all so much for your time.
- 1:34:22
- Unidentified speaker (UNKNOWN): Thanks
- 1:34:22
- Unidentified speaker (speaker_0): everyone
- 1:34:23
- Unidentified speaker (speaker_3): for that, Sam.
- 1:34:25
- Unidentified speaker (speaker_2): Thank you for listening to this episode. of FYI, the For Your Innovation podcast. If you enjoyed this episode, please hit that like button and make sure to leave a comment down below letting us know what you thought. And if you haven't already, please make sure to subscribe to the show and follow us at Arkinvest on your favorite social platforms so you stay up to date whenever we
- 1:34:44
- Unidentified speaker (speaker_2): drop a new episode.
- 1:34:50
- Unidentified speaker (speaker_2): ARK believes that the information presented is accurate and was obtained from sources that ARK believes to be reliable. However, ARK does not guarantee the accuracy or completeness of any information, and such information may be subject to change without notice from ARC. Historical results are not indications of future results. Certain of the statements contained in this podcast may be statements of future expectations and other
- 1:35:10
- Unidentified speaker (speaker_2): forward-looking statements that are based on ARK's ARC's current views and assumptions, and involve known and unknown risks and uncertainties that could cause actual results, performance, or events to differ materially from those expressed or implied in such statements.
More from FYI - For Your Innovation
Filling The Falcon 9-Shaped Hole: Stoke Space's Bet On Full Reuse
Episode discussion: Stoke Space VP Devin Papandrew discusses full reusability, Nova Block 2, and filling the Falcon 9 gap
Speakers’ claims: Devin Papandrew, VP of Business Development at Stoke Space, explains the company's strategy to build 100% reusable rockets using actively cooled engines to solve re-entry heat challenges. He argues that vertical integration and full rapid reuse are essential for controlling costs and achieving the high cadence needed to compete with SpaceX. Papandrew highlights the Nova Block 2 rocket, designed to capture demand left by SpaceX's capacity constraints, and identifies microgravity manufacturing and defense as key growth drivers. The discussion covers technical milestones, including over 11,000 seconds of hotfire testing, and targets a first flight in early 2027.
Did The Midterms Just Break The AI Trade? | The Brainstorm 157
Episode discussion: Midterm election impacts AI regulation and compute economics, favoring infrastructure owners over frontier labs
Speakers’ claims: Speakers argue that midterm election outcomes may delay AI data center construction by ~10% and increase federal regulation, potentially benefiting incumbents like Meta and Alphabet by giving them time to catch up to frontier labs like Anthropic and OpenAI. They contend that compute scarcity allows infrastructure owners to charge premium rates for renting compute, making it economically irrational for frontier labs to invest in competing models. Anthropic's growth is attributed to product-led adoption by engineering teams via API tokens, with high margins supporting a rumored $1.1 trillion valuation. The discussion also covers risks associated with open-weight models, including security vulnerabilities and benchmark gaming, while debating the future monetization of AI in search and advertising.