Nucleus Investment Insights

NVIDIA's valuation reflects margin normalization risks amid leadership transition and competitive pressure

41 min Listen to the original Episode page
NVIDIAAI InfrastructureValuationLeadership TransitionCapExSemiconductorsGeopolitics

Summary

Damian Classen argues that NVIDIA's low valuation multiples (7-10x forward earnings) price in a significant decline in market share from over 90% to 60-70% and margin compression following Jensen Huang's departure. He contends that Huang's aggressive supply chain management and startup-style leadership are difficult to replicate, suggesting the next CEO will likely be more conservative. Classen dismisses concerns about neo-cloud debt markets, asserting that hyperscalers will absorb any construction projects, keeping AI CapEx robust. He identifies inference demand, rather than training spending, as the true indicator of the AI boom's sustainability, comparing the current cycle favorably to the dot-com era due to its focus on physical infrastructure.

Key takeaways

  1. 1
    CoreWeave's potential funding failure would not stop data center construction; instead, hyperscalers like Microsoft or OpenAI would likely build the infrastructure, shifting capital from neo-clouds to established tech giants.

    Damian Classen · 29:55

Episode sections

  1. 0:30 – 2:52

    Introduction and NVIDIA Leadership Transition

    Damian Classen introduces the episode, noting NVIDIA's former status as the market's biggest stock and the significance of Jensen Huang's departure. He outlines three key discussion points: NVIDIA's specific position in the AI infrastructure market, the geopolitical/political implications, and the broader lessons for the AI boom.

  2. 2:52 – 5:37

    NVIDIA Financials and Valuation Analysis

    Classen argues that recent strong earnings results prove little given Huang's short absence. He highlights NVIDIA's valuation as 'dirt cheap' at 10x next year's earnings and 7x the year after, compared to the market's 18x multiple, despite 100% earnings growth. He questions the sustainability of these margins and market share.

  3. 5:37 – 8:47

    Competitive Landscape and Market Share Forecast

    The speaker describes NVIDIA's position as having a 'long lead' via its GPU and data center operating system, but notes intense competition from well-financed customers like AWS, Amazon, Google, AMD, and Intel. He predicts NVIDIA's market share will fade from 90%+ to 60-70% over 5-10 years, though it will remain the sector leader.

  4. 8:47 – 11:24

    Margin Sustainability and Normalization Scenario

    Classen suggests NVIDIA's current margins are inflated by scarcity. He posits a scenario where margins halve but volumes double, resulting in similar or higher absolute earnings. He argues this 'normal semiconductor' outcome is largely priced in, and further upside depends on cash flow usage like buybacks.

  5. 11:24 – 13:33

    Jensen Huang's Management Style and Succession Risk

    Huang is described as running NVIDIA like a startup with 60 direct reports, a structure difficult to replicate. Classen compares the succession risk to Apple post-Jobs, suggesting the next CEO may manage rather than innovate, potentially leading to slower growth and faster margin erosion.

  6. 13:33 – 16:43

    Supply Chain Management and Financing Aggressiveness

    Huang is credited with exceptional management of memory shortages by locking in low rates with Micron, Samsung, and SK Hynix. Classen notes Huang's aggressive financing strategies may have accelerated growth beyond sustainable levels, and the next CEO is likely to be more conservative.

  7. 16:43 – 25:35

    Geopolitical Scandal and Sanctions Scenarios

    Classen discusses the political fallout of Huang's scandal, outlining three scenarios: a political hit job, a genuine push for stricter sanctions by China hawks, or a failed attempt with no real change. He suspects Trump's administration prefers 'release valves' in enforcement, allowing China to skirt sanctions while maintaining headlines.

  8. 25:35 – 29:36

    Impact on AI Capex and Data Center Construction

    The speaker analyzes factors affecting AI Capital Expenditure, focusing on Texas state politics. He argues that political rhetoric regarding water usage and grid connection is largely meaningless because data centers are already building their own power and recycling systems. He expects little regulatory slowdown from state legislators.

  9. 29:36 – 30:41

    Neo-cloud debt markets and CapEx impact

    Damian Classen argues that neo-clouds like CoreWeave are minor players compared to major loss-making AI firms like Anthropic, OpenAI, and SpaceX. He suggests that if neo-clouds face debt market closures, their data center projects will likely be built by hyperscalers instead, meaning the broader AI CapEx boom will continue regardless.

  10. 30:41 – 32:02

    CapEx slowdown effects on NVIDIA vs Hyperscalers

    Classen posits that a slowdown in CapEx growth is negative for chip sellers like NVIDIA and Broadcom but positive for hyperscalers. Slower supply combined with rising demand allows hyperscalers to charge higher prices for capacity, benefiting them while hurting equipment vendors.

  11. 32:02 – 34:40

    AI Training demand and the 'arms race'

    The speaker identifies AI training demand as driven by an 'arms race' among companies like Anthropic, OpenAI, Google, Meta, and SpaceX to build the best models. This competition forces continuous overspending on R&D, underpinned by open capital markets, keeping GPU rental prices high.

  12. 34:40 – 35:00

    Inference demand as the true cycle indicator

    Classen asserts that AI inference (end-user usage) is more critical than training demand for sustaining the boom. Even if training spending stalls, the cycle can continue if end-user demand remains strong, driven by the broader economy's willingness to spend on AI queries.

  13. 35:00 – 36:02

    Token prices vs. GPU cost per hour

    The speaker advises investors to ignore falling token prices and rising token volumes, arguing that the key metric is GPU cost per hour. This metric reflects actual spending, as increased volume can offset lower per-token costs, maintaining total expenditure.

  14. 36:02 – 40:33

    Comparison to the dot-com boom

    Classen contrasts the current AI boom with the dot-com era, noting that early-stage internet spending was sustainable, but the peak involved companies raising money to spend on each other's advertising. In contrast, current AI spending is directed toward building physical data center capacity for genuine end-user queries.

Assets discussed

AssetKindStanceContext
NVIDIANVDAequitymixedthis week's show. Um, NVIDIA. So the empty throne, the uh of the AI boom and uh Jensen Huang.
AmazonAMZNequityneutralAmazon, Microsoft. And at the moment, they are completely open. Look, the rates they're paying have gone up and then gone up a lot, I guess, relative to government bonds.
MicrosoftMSFTequityneutralAmazon, Microsoft. And at the moment, they are completely open. Look, the rates they're paying have gone up and then gone up a lot, I guess, relative to government bonds. But these things all had net cash. And so you sort of basically gone from, okay, they've they've gone from being priced almost like a government to now they're being priced more like a company. There hasn't been a significant change. If Google wants to lend money, there is no problems with people lending money to Google. Same with Microsoft.
AlphabetGOOGLequityneutralIf Google wants to lend money, there is no problems with people lending money to Google. Same with Microsoft.
Meta PlatformsMETAequityneutralThe next part is equity markets to places like Anthropic, SpaceX, OpenAI are your big ones. But also they meta did a capital raising recently and things like that.

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

Jensen Huang has been absent from his role for only one week as of the recording.

Unidentified speaker (UNKNOWN) ·3:00

How to verify: Check the date of Jensen Huang's departure announcement and compare with the podcast release date of 2026-11-19. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

NVIDIA trades at 10 times next year's earnings and 7 times the year after.

Damian Classen ·4:48

How to verify: Calculate NVIDIA's forward P/E ratio using consensus analyst estimates for fiscal years 2026 and 2026 and compare with the current stock price. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

The broader market trades at approximately 18 times earnings.

Damian Classen ·5:07

How to verify: Check the current P/E ratio of the S&P 500 index. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

NVIDIA has over 90% market share in its sector.

Damian Classen ·7:17

How to verify: Review market share reports from semiconductor research firms like Gartner, IDC, or TrendForce for AI accelerators/GPUs. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Jensen Huang has 60 direct reports.

Damian Classen ·11:44

How to verify: Check NVIDIA's proxy statement (DEF 14A) for the organizational structure and reporting lines of executive officers. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

NVIDIA secured favorable memory rates with Micron, Samsung, and SK Hynix during a shortage.

Damian Classen ·13:53

How to verify: Review NVIDIA's earnings call transcripts and SEC filings for discussions on supply chain management and cost of goods sold during the relevant period. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Texas state legislators are considering regulations on data center water usage and grid connections.

Unidentified speaker (UNKNOWN) ·27:00

How to verify: Search Texas legislative session records and public utility commission filings for proposed bills or regulations regarding data center infrastructure. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Google and Microsoft have open debt markets and are net cash positive.

Damian Classen ·28:48

How to verify: Check the latest quarterly financial statements (10-Q) for Alphabet (Google) and Microsoft to verify cash positions and debt issuance activity. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Meta recently raised capital.

Damian Classen ·29:15

How to verify: Review Meta Platforms' SEC filings for recent equity offerings or debt issuances in the 2026-08-19 to 2026-11-19 period. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

China is skirting sanctions to purchase NVIDIA chips or rent compute power.

Damian Classen ·18:10

How to verify: Review US Department of Commerce Bureau of Industry and Security (BIS) enforcement actions and reports on semiconductor exports to China. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic, OpenAI, and SpaceX are major loss-making companies driving AI infrastructure demand.

Damian Classen ·29:36

How to verify: Check SEC filings (10-K/10-Q) for public subsidiaries or parent companies (e.g., Microsoft for OpenAI stakes) and press releases regarding their financial status and CapEx commitments. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

CoreWeave is facing debt market issues and is a 'bit player' compared to hyperscalers.

Damian Classen ·29:55

How to verify: Review CoreWeave's recent financing announcements, credit rating reports, or news articles from 2026-08-19 to 2026-11-19 regarding its capital structure and market share. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

If CoreWeave fails to fund data centers, hyperscalers like Microsoft or OpenAI will build them instead.

Damian Classen ·30:16

How to verify: Monitor press releases from Microsoft, OpenAI, and CoreWeave regarding data center construction contracts and funding sources during the verification period. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

A slowdown in CapEx growth is negative for NVIDIA and Broadcom but positive for hyperscalers.

Damian Classen ·30:41

How to verify: Analyze NVIDIA and Broadcom earnings reports for revenue correlation with CapEx cycles, and review hyperscaler earnings calls for margin improvements due to pricing power. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic is listing 'tomorrow' (relative to the transcript date of 2026-11-19, implying a late November 2026 IPO).

Damian Classen ·33:41

How to verify: Check SEC EDGAR for Anthropic's S-1 filing and NASDAQ/NYSE listing announcements for late November 2026. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

GPU rental companies can secure 20-40% of their revenue from AI model training workloads.

Damian Classen ·34:01

How to verify: Review financial disclosures of major cloud providers or GPU rental firms (e.g., CoreWeave, if public, or hyperscaler segments) for revenue breakdown by workload type. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Token prices are falling exponentially while the number of tokens produced is increasing exponentially.

Damian Classen ·35:00

How to verify: Access public API pricing data from major AI providers (OpenAI, Anthropic) and aggregate usage statistics from industry reports (e.g., from AWS, Azure, or independent AI analytics firms) for the period 2026-08-19 to 2026-11-19. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

During the dot-com boom, companies like Cisco were priced at 100 times forward earnings.

Damian Classen ·37:59

How to verify: Historical financial data from Yahoo Finance or SEC filings for Cisco Systems (CSCO) during the 2026-2026 period to verify valuation multiples. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

NVIDIA is currently priced at 10 times earnings for next year and 7 times earnings for the year after.

Unidentified speaker (UNKNOWN) ·37:39

How to verify: Check NVIDIA's (NVDA) stock price and consensus analyst estimates for forward P/E ratios as of 2026-11-19 via financial data providers like Bloomberg or FactSet. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

The AI boom is driven by spending on data centers for end-user queries, unlike the dot-com boom's circular ad spending.

Unidentified speaker (UNKNOWN) ·36:31

How to verify: Compare historical SEC filings of dot-com era companies (e.g., Pets.com) with current AI company CapEx reports to distinguish between operational infrastructure investment and marketing/advertising spend. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Do your own research

  • Verify NVIDIA's current forward P/E ratio and compare it to the S&P 500 average using a financial data provider like Bloomberg or Yahoo Finance.
  • Review NVIDIA's latest 10-K and 10-Q filings for details on gross margin trends and segment revenue breakdowns.
  • Analyze competitor announcements from AMD, Intel, and hyperscalers (AWS, Google, Microsoft) regarding their own AI chip development and data center operating systems.
  • Check NVIDIA's insider trading activity and succession plan announcements following Jensen Huang's departure.
  • Monitor GPU rental pricing trends from major cloud providers and neo-clouds like CoreWeave to assess the 'GPU cost per hour' metric.
  • Review Texas state legislative records and utility company filings to assess actual regulatory hurdles for data center power and water usage.
  • Examine the debt financing structures and recent bond issuance of neo-cloud providers to evaluate their liquidity risks.
  • Compare current AI CapEx trends and infrastructure spending with historical dot-com era advertising spend data to validate the 'physical infrastructure' vs 'circular spending' argument.
Full transcriptexpand

0:30Unidentified speaker (UNKNOWN): Hello,

0:31Damian Classen: welcome to Nucleus Investments Insights Podcast. Each week we explore the trends shaping stock markets, currencies, commodities, global investment themes, giving you the insights you need to stay ahead. My name is Damian Classen, Chief Investment Officer at Nucles Wealth, be hosting and sharing my insights and findings throughout the show. Don't forget to like and subscribe if you find our insights useful and you want more of them in your feed. Now on to this week's show. Um, NVIDIA. So the empty throne, the uh of the AI boom and uh Jensen Huang. Um, so obviously, you know, biggest stock in the market, or formerly the biggest stock in the market, um, losing its CEO uh was a pretty big story and pretty important for a whole bunch of different reasons around uh around the AI. Um, just a quick reminder that everything you hear today is. general advice in nature, none of this personal advice. If you do want personal advice, you can follow the show notes, um, speak to uh one of our team to help you with it. Uh right, so back to

1:26Unidentified speaker (UNKNOWN): uh

1:26Damian Classen: NVIDIA. And I think the the key things I want to sort of run through, there's there's a couple of different things that really stand out in terms of this. So it's a really big stock and it's a it's a major part of the indexes, uh, major part of our portfolio. It's been for for a long time as well. Uh, and so there's questions about Nvidia and its position in the whole AI um you know infrastructure and and and and market within within that sector, um, which is important and about its growth rates and and things like that. That's That is well that's worth exploring. There's also

2:04Unidentified speaker (UNKNOWN): a

2:05Damian Classen: second issue, um, which is the political side of it and what what this is sort of what this whole thing is saying about the geopolitical um sides of it. And then the final part I want to focus on is sort of the the lessons within this for the AI boom. So uh yeah, so three different stages all sort of matter in their in their own. and they'll obviously interlink with each other. But I think it's well worth thinking about these as separate issues that are important, sort of coming to because

2:36Unidentified speaker (UNKNOWN): all

2:36Damian Classen: three of them will make a difference on a whole range of other investments. Um uh and so anyway, so let's let's talk about NVIDIA separately as an as an individual company and and what some of the issues are. And so it's it's just reported that there's results, results. look great. The thing is it sort of proves nothing. You know, and Jensen hasn't been there for a week. And so

3:00Unidentified speaker (UNKNOWN): he

3:00Damian Classen: can't prove that that everything's fine using this result. It was, you know, vaguely possible that maybe you'd find there was a, you know, you could have disproved it um that everything was fine by saying that a whole rash of cancelled orders and and and an announcement said they're they're making wholesale changes to their funding and things like that. But really, my expectation is that what we'll see is an interim CEO who's basically not going to do much at all, uh just going to try and keep the uh the plane in the air. And then uh you'll get a new CEO at some point. You know, there's some rumors bouncing around, at least that Jensen's trying to get back in. Um, but you'll end up with another CEO. and the question about what they do and where they see where they sit is um, you know, is I guess open to to it. But it's probably more likely you're gonna end up with somebody uh sort of similar to Apple situation where you have a visionary founder who creates a bunch of things, including say the biggest one, the uh the iPhone, and and then once he leaves, the next guy comes in. and really just manages

4:04um that and sort of it's it's a question about keeping that going rather than trying um rather than creating all these brand new things out there. And I think that's probably uh the case you'd sort of look at for for NVIDIA. There's just lots of other things going on behind that in terms of uh where they are in their cycle, which is just different. Um, and and worth sort of we'll discuss that in in a in a minute, uh, about how that fits together.

4:27Unidentified speaker (UNKNOWN): Um,

4:28Damian Classen: so first, so the result uh basically saying, look, everything looks great, but doesn't really make any difference uh as to uh you know, it doesn't really tell us anything about about what's really going on within this. The second part is then saying, um, you know, valuation-wise, how does it look? Uh look, it's cheap. It's you know, 10 times next year's earnings, it's seven times. a year after that's earnings. Um, it's a uh and and market the market's sort of trading on sort of 18 times. So you're sort of half the price of the market for this, not not quite, but almost half the price of the market, uh that the average stock in the market, and it's growing at sort of you know 100% in terms of its earnings. So uh yeah, from from a from a from a

5:15Unidentified speaker (UNKNOWN): yeah

5:16Damian Classen: take a big step back this thing just looks dirt cheap um and uh you know sort of a bit like a no brainer in terms of the uh in in terms of where it's sitting but then the question comes about about sustainability of earnings and and that's probably the key factor to this whole um putting it all together so

5:37Unidentified speaker (UNKNOWN): I

5:38Damian Classen: think the broad view on an NVIDIA, which I think is well, the broad view we've got internally, and you know, you obviously see

5:45Unidentified speaker (UNKNOWN): um

5:46Damian Classen: different opinions from different cases, but was that look, they've got a long lead over a lot of their competitors, you know, entrenched software, uh, as well as um yes, I've got this, they're an operating system. It's not just the GPUs. They they went from having you know the best GPUs out of the market, leveraging that into an operating system, and leveraging that into effectively being um you know the operating system for data centers that come out. You think about it as, you know, if you're buying a computer and you're putting Microsoft Office on it, it's a bit like that. You're buying a data center, you're putting a lot of the NVIDIA chips, NVIDIA for operating systems, uh, and it makes everything talk within that. So So so so from that perspective, yeah, the the the view is they've got a long lead. but they've got a lot of competitors who are uh well financed, you know, some of the biggest companies in the world that are busy chasing

6:36Unidentified speaker (UNKNOWN): them.

6:37Damian Classen: And and most of these are the some of their biggest customers. So whether it's um you know uh AWS or Amazon and Google and and and other companies like that, that have all got their own chips um that are out there, there's a whole bunch of other stuff. you know former leaders of the industry or or coming or leaders in other other markets, sort of uh AMD or or Intel and and and companies like that that are all trying to break into this as well. Um and you know, while they're not nowhere near as big, that they are that they're still big and they've got big R and D departments. And so, you know, there's a chance that one of these comes up with a breakthrough that that that you know leapfrost. leapfrogs NVIDIA where they are. So yeah, so potential, but but you

7:18Unidentified speaker (UNKNOWN): know

7:18Damian Classen: that that's the case in any in any sector. So so the big picture view on NVIDIA is, yeah, ninety percent plus market share, it's gonna fade over time. I think the uh you know, do they end up with a sort of a sixty, seventy percent is probably the um, you know, w without seeing any any you know any major competitors with with genuine things popping up is is it seems like a reasonable assumption that that over time over the next sort of five to ten years they'll end up with a a much smaller part of uh of that but still being the number one uh company in in the sector.

7:51Unidentified speaker (UNKNOWN): Uh

7:51Damian Classen: and and that sector was probably never going to grow to being hundreds of competitors within that. So uh most of the semiconductors that we look at uh tend to end up with with a quite a small number of you know three to sort of five major players within that who who have the the biggest market shares. And then after that, there's there's a number of sort of also rans within that. And sometimes you get you know switches around. So so yeah, so if you sort of took the view that look, um they've got they're sitting in this market, they're not gonna be able to keep their market share, but the market's gonna grow itself. So so maybe you you you're possibly looking at

8:26Unidentified speaker (UNKNOWN): sort

8:26Damian Classen: of stuff.

8:27Unidentified speaker (UNKNOWN): uh

8:27Damian Classen: similar revenues, I suppose. Or so yeah, similar dollars coming in. So you go, yes, I go from having a 90% market share to a 60% market share, but um the you know it's just it's just increased in 50% in size. And so the dollars is actually better off anyway, a smaller part, but a bigger slice, so to speak, um, in terms of absolute volume. So there's that within NVIDIA. Um and then the question then is is the margins that they're earning. They're they're definitely earning um, you know, margins that are way over the top at the moment, uh, because of the scarcity issues. Having said that, they don't produce these chips themselves. And so this is more of a pricing, there's definitely a demand supply thing here where

9:09Unidentified speaker (UNKNOWN): um,

9:09Damian Classen: you know, they they could halve the value of their their their chips and you would they'd still be making pretty reasonable margins for it. So so you know, the this whole question then sort of comes about saying, okay, so if we look forward at the next year's earnings, that's pretty locked in. You know, there these orders are already in for for for those. It's probab a little bit of upside or downside, but you know, you're not gonna be off by orders of magnitude in terms of that. The second year out is probably looks pretty good as well. Um, most of that's uh, you know, these are for data centers that are being built. They've got applications in, you know, they're working their way through the process. Uh, you know, there's the next year to two years is probably relatively locked in.

9:51Unidentified speaker (UNKNOWN): And so

9:51Damian Classen: you look at a company on seven times earnings on that. And then the question is saying, well, what if their earnings, what if their margins halved um uh and their sales increased though, because it it uh they're gonna start getting into the replacement cycle for some of these things as well at some point. And it looks like data centers are just gonna be bigger and and more here too. So, you know, even if you've gone through and said, Okay, so the the the their margins literally halve the price they sell these things for, um, but their volumes go up and and net effect is that they earn earn the same amount of money or or maybe a little bit more than that, you're probably looking at at a at a semiconductor at that point. So if you're buying a you're buying NVIDIA now and basically saying, look, it's all going to evaporate over the next couple of years, you're just going to be left with a normal semiconductor stock in say three to five years with a sixty percent market share and and half the margins is what they they have at the moment, then

10:43Unidentified speaker (UNKNOWN): um, you know,

10:44Damian Classen: that's pretty much what's priced in already uh in terms of stocks. And so any extra growth they get, any any

10:52Unidentified speaker (UNKNOWN): uh

10:52Damian Classen: the support of the extra cash flow, what they do with that extra cash flow, uh any of the you know the buybacks and stuff like that that's going on, I think that looks pretty good. So from that perspective. I guess what I'm saying is um there's a uh your downside needs to come from from suggesting that this is gonna To unravel um the larger AI trade, and there's other things going on within that. Okay, so that's sort of NVIDIA. um, big picture wise in terms of valuation. The other thing I want to talk about then is just the is Jensen's role within that, and it's significant. So it's not and and it's significant in a way that you don't see in a lot of companies. It really does seem to be a major company uh run like a startup. So he has um sort of 60 direct people reporting. to him, which is sort of unheard of for a company of that size. And his view was well, these are all senior people, they all know what they're doing. I don't need one on one conversations with them on a daily basis. You know, it's more about the empowerment, so to speak, of some of these um lower reports. But it's also about so in some ways it's that, but it is also about that uh, you know, he's really is directly running that company. like a startup. He's a founder.

12:04When he decides to change direction, uh he's got all these direct reports. It's not a um it's sort of straight through. And so uh are you gonna be able to find somebody that fits straight into that that chair? It's very unlikely. It doesn't seem to be that many people internally, even that are going to that seem to be set up for that. Like you know, he he did a lot of the fronting uh and left uh you know a lot of the uh the next sort of layer of management really got left to to to do the operating without having to worry about the whole financing and speaking to investors, that type of stuff like that. So uh yeah, there's definitely they're big shoes to fill. And the question is, you know, do you end up with a company that uh sort of goes through that Apple case uh scenario where okay the next person in there is now just gonna he's is gonna run a much more traditional company. It's much more likely to run like a a normal one. And much more likely to run, yeah, like a normal company, but which means it's probably not gonna grow as fast. Uh you're probably gonna miss out on some of the you might not be able to keep the uh the margins that you have for as long. So so those I spoke about it and the margins coming off and and losing market share, maybe that happens faster uh under somebody else, or quite likely happens faster.

13:15Unidentified speaker (UNKNOWN): Uh

13:15Damian Classen: and then the other thing is within this is is that you look at

13:20Unidentified speaker (UNKNOWN): um

13:20Damian Classen: there's certainly he he Jensen has managed to have his finger on the pulse and and make changes at a significant level, um, and and pick the right winners.

13:33Unidentified speaker (UNKNOWN): So to

13:33Damian Classen: speak. Not say everything's a winner. Winner certainly he's got he's had a share of losses as as everyone does in in in these types of. setup companies. But one of the things he has been focused on is things like say uh we ran into a memory shortage over the last year, which we've seen, you know, probably seen a lot about this whole Micron, Samsung, SK hinex prices, you know, uh going through the roof and those companies making all this money from it, is that uh he was onto that really early and went and struck a deal with all of them to to to get his memory at much lower. rates and locked it locked in these rates and and and really um yeah his management of that um was exceptional and and and in no small way, you know, uh showed how he his finger on the pulse and it meant that the you know he kept his profitability and and managed sort of drive that forward. So uh yeah, so I think uh it it's it's very hard to see him being replaced by somebody who's gonna be you know in that same position. So you I think that lower growth, um, you know, more cash flow returns for shareholders

14:33uh is important. Um and uh so is that that's one part. I will talk a little bit about the financing side because there was certainly uh you know the question in terms of what he's been doing on the financing

14:44Unidentified speaker (UNKNOWN): um

14:45Damian Classen: is he's been quite aggressive in terms of in terms of the financing, which sort of led to a bigger market than than it might have otherwise been.

14:53Unidentified speaker (UNKNOWN): Uh,

14:53Damian Classen: that's likely to go away. I would think the next CEO is going to be much more conservative on that, on that front. They certainly do some of it and and it's unlikely to stop too much of the existing ones, but there was certainly um, you know, a speed of growth. Now, that might not that may or may not be a a a good thing or a bad thing. So in in some ways, there's this question about, and this is before I get to the China exposure, but it's sort of related in a way. There's this question about saying, well, when he's selling, he's got more chips. Sorry, he can't sell enough chips out there. The demand for his chips is is is greater than than uh the ability to supply and and these data centers to get up and running, is that uh in a way he's sort of picking and choosing. And so you go, well, so maybe there's some neo cloud that doesn't get up now and doesn't manage to that might have otherwise managed to get a deal with Jensen, but they don't get a deal, and so they don't get to to to buy chips from him, uh, because of the financing and all that type of stuff like that. Um, but maybe those chips just end up somewhere else at a different neo cloud or one of the the hyperscalers. And that's sort of similar with the China bit, is saying, okay, if China can't buy the, you know, skirt the regulations and

16:04buy the chips that they've been buying, or effectively buy the chips by by renting them, does that just mean that those end up somewhere else? And there's a reasonable argument for yes at the moment. Um that's the case. Uh the question is, you know, does it mean that this that extends the boom out longer for others? And then there's a pretty reasonable scenario in terms of that case.

16:24Unidentified speaker (UNKNOWN): Um uh

16:25Damian Classen: and then so that whole part about um you know demand demand supply uh is is really hard to tell what's going on within that. Uh okay, so I might leave NVIDIA um there for the moment. And that's I want to because I want to talk about the politics, the two second parts of it, the politics of

16:43Unidentified speaker (UNKNOWN): it, and

16:44Damian Classen: then the lessons for for the the broader AI sector that we really need to to think about. So the big picture is this is in in in some ways, sorry, in many ways, yeah, it's a it's a tawdry sex story which people love and and make gets lots of headlines and one of the richest guys in the world and all this and Chinese spies and you know it's all it's got all the intrigue from the front page of the newspaper. But I wouldn't go past the whole political side of it. So it's come as pr as a leak from security apparatus within the US. So it's clearly been designed to get it out there. It's been known for some time. And so there's a whole bunch of things that that sort of comes into as about okay, well, what does this mean? What are the different um places we can come to from here? Now, at one point, um, you just go, Oh, look, it's just a political hit job aimed at Republicans. Yes, yes, NVIDIA goes. got hit, but that was just crossfire. This was more about saying whoever Trump knew about this or Lutnik knew about this and didn't do anything about it. So that's that's one and it's just a political hit job and and you know that's that's as far as it

17:50goes. Um put that aside for a minute, because I'm gonna go through the different scenarios and then we'll come back to this. Secondly, is um could this be a genuine attempt by the the China hawks within the administration and and you know the lower levels to to get stricter sanctions. That they they can see that uh China's you know getting around the thank sanctions or or not completely, but you know, um skirting the s sanctions and and it's at at the edges.

18:15Unidentified speaker (UNKNOWN): Uh

18:15Damian Classen: and that's gonna lead to a much stricter sanctions regime. Uh and then the other one is it's it's a genuine attempt by China hawks, um, but they haven't got the political power to actually make it get to get it to strict just. stricter sanctions. Yeah, it's gonna be it'll blow up or whatever, but it's not gonna actually end up in in much of a difference. And so um yeah, three different three big scenarios. Uh and

18:38Unidentified speaker (UNKNOWN): I

18:38Damian Classen: think the thing uh I sort of want to discuss around this is that look Trump has obviously been uh quite negative from a big picture point of view on terms of China. So he's got his tariffs, he's got his bans and and all this stuff going on within that. So he loves the headline in terms of China's evil, uh, it's uh I'm gonna I'm gonna get China, they've been screwing us for years, and and you know, that that um that playbook. But he's never really been big on that whole enforcement and certainly the minor enforcement around the edges. And so the question is: how much of this is release valves that are there that are, you know, say on the tariffs or or on these chip bands and stuff like that, which is like, well, it's just accepted. It's a, yeah, we're going to put a ban on this, but we know it's not going to be a hundred, nothing's going to be a hundred percent. And actually, sometimes it is better to have a bit of a release valve and you know, stuff at the edges gets through. How much of that is sort of intentional and and whether they're getting paid in in various ways for that happening, and how much of that is is sort. of more

19:39that he doesn't care about the detail. So I'm just gonna go to one quick message and then let's come back and talk about this. Uh right. Sorry, I've got to think I've got a technical problem on those. Okay, so uh yeah, let's talk about the politics of it. So

19:51Unidentified speaker (UNKNOWN): um

19:53Damian Classen: I mean, look, I don't know. Nobody really knows w within this. Uh we'll probably never know as as to what's uh as to what it is. I do think that that it it is worth thinking. about Trump on those from that perspective of uh he wants the big picture announcements of I'm banning this, I'm tariffing that, I'm doing whatever it is. Uh you know, I'm the strong man. But I don't think whether it's a lack of um detail that he cares about or whether it's a um an intentional sort of leaving these safety valves, I think that's been pretty much his modus operando on on a whole bunch of things. So so from his perspective. I suspect that's exactly where um exactly where this is headed from his perspective. Now, um, you know, this the I guess the the question of the first question I raised is is this a political hit job just aimed at Republicans? Which is, you know, completely possible. Uh and and NVIDIA is just a bit of a they got hit in the the the crossfire. Um now, where does that go though? Because that sort of doesn't, in a way, it doesn't really matter whether whether it was or it wasn't. Um the bigger issue is

20:53does this lead to stricter sanctions or not? So does it end up in the path of like, you know, people are so outraged by this that it has to end up in in stricter sanctions, or d is it something um which doesn't? And and of there's a the flow-on effects of this are important, and they're probably almost in a way um a little bit uh cot contra um intuitive. in terms of in terms of what I think, in terms of that, the the broader AI, what it means for

21:19Unidentified speaker (UNKNOWN): it.

21:20Damian Classen: So

21:21Unidentified speaker (UNKNOWN): um I

21:22Damian Classen: don't know where it's gonna go. My my guess is it probably doesn't it probably doesn't lead to hugely stronger sanctions. It probably leads to um you know some some big announcements about the things that are gonna be happening or whatever it is. Um, but you know, it does probably doesn't lead to to much difference in terms of how things have been. are being done. And China will skirt around this, and there'll be some new sanctions or new whatever, and China will skirt around them. Um, or maybe maybe they won't, maybe it'll just end up as a as a as a talk face and and and complaining and how to uh how to get rid of Lutnik and and

21:54Unidentified speaker (UNKNOWN): or

21:55Damian Classen: pick whoever you you they they're going to gonna aim at in in terms of the the the the regime in terms of the um you know Trump administration. So uh yeah so so why this matters that's a bigger part so why it matters is uh US companies actually do all right out of out of China on this on the on the current setup is that okay so China at the margin can skirt um what does that mean? Well it actually means that for the large part China's

22:24Unidentified speaker (UNKNOWN): you know

22:25Damian Classen: putting money back into US companies' hands. which is you know it's it's whether it's paying for the its chips through there, it's not using its own chips, um it's using some of the frontier models. And and I guess in a way the US companies can see what China's actually doing, and there's a lot more openness. China's very much funding and um encouraging the open source sector. And so what it is doing from that perspective is um it is diversifying the the customers. And in a way, that's sort of the question for NVIDIA as well is about saying, well, maybe if they didn't sell these chips to China or if they didn't sell these chips to the Neo Clouds of the funding, that those chips just would have been bought by somebody else anyway. But it would have meant that Microsoft gets more or Google gets more. And so they still sell the same amount, but then they've got a much more concentrated um user base. And so maybe they're better off with with actually a more spread out and more of these new. neo clouds and more of the open source sector. So I do think if it does end up being a big bit of a crackdown, that's probably a negative for that part of the market, which it which will net in net be a negative for NVIDIA. Um, but it's on the same side it's probably a positive for the for the bigger hyperscalers. So, and that's certainly how we're looking at our portfolio is we've got us hyperscalers and we've got NVIDIA. NVIDIA makes money, that's a that's a uh that's a cost for. these hyperscalers. Um NVIDIA's margins shrink, well, that's now a benefit to the

23:43hyperscalers. And so it's a little bit of a as long as you're as long as that the whole uh AI boom doesn't come unraveled, it's a little bit of an H-way bet to say, okay, I want some NVIDIA for that part, but I also want the hyperscalers for the other part to say, you know, one of them's gonna win out of this. But but the whole China, this whole China part is if you saw some sort of significant banner. ban on this, then it's probably a negative for the the AI sector as a whole in in the longer term, in that more likely to see more concentration, so less of the open source side, and more likely to see that China's jet gonna have to um create its own chips. And so I mean it's obviously trying to at the moment but it's um trying to but then using the the the US ones wherever it can And the thing is, well, will it get to the stage where it's um if it's completely shut out of the market, then it has it, it's it's it ends up being a lot more focused on the on generating its own chips and eventually maybe it gets somewhere in terms of um in terms of where that goes. And and then the market's just smaller for the US companies.

24:45Unidentified speaker (speaker_1): Okay.

24:46Damian Classen: So yeah, where that comes out, I think it will make a difference. I do think it's very much at the margin. Uh uh, but you know, I think the answer here is either

24:55Unidentified speaker (UNKNOWN): um,

24:56Damian Classen: yeah, lots of fire and lots of brimstone and and not much really changes and and maybe there's a few roadblocks put up that that China just go a different way around and and everything's roughly the same. Um, or there's a genuine um shutdown. Down. So that's that's one of the factors watching. What's gonna happen? I yeah, don't know. It could could be either.

25:14Unidentified speaker (UNKNOWN): Um

25:15Damian Classen: the next question then is the broader AI boom. So so just keeping this in mind again. The first part I want to talk about was NVIDIA and and what's been happening there as an individual company. Secondly, I think the big thing is this geopolitical issue and where that actually goes to. And the final one is um the AI boom itself. And and I think there's four things we really need to focus on in this part. Um is one is the AI capex side. Uh, and then oh sorry, this there's two parts to this. There's so four parts, but in two big chunks. One is the AI Capex phase, and the other one is the AI actual use of AI.

25:52Unidentified speaker (UNKNOWN): And

25:52Damian Classen: so on the CapEx side, I think there's two major issues that we're looking at from this. Um, one is will political forces slow down AI CapEx? And this is more. at the state level, I think, than than the federal level. And this is where I think Texas is probably your key one here.

26:09Unidentified speaker (UNKNOWN): So

26:09Damian Classen: they've just changed their state legislature, heaps of uh development of uh AI data centers in Texas, got cheap energy, lots of land, sort of fits um and and um very open business regulations. regulation. So uh that's where a lot of the the boom has sort of been focused. Now the question is, will the political forces actually come in to slow that down?

26:33Unidentified speaker (UNKNOWN): I

26:33Damian Classen: don't think the ones we're looking at, uh, I don't think what we're seeing from video at the at the top level is going to do that from a federal level.

26:41Unidentified speaker (UNKNOWN): Uh

26:41Damian Classen: the question now is with the state legislators, and that's really what we're watching within that. I I suspect, um, but I obviously don't know. I suspect that what we're going to see is largely what we've seen from other ones, which is lots of noise, actually not much in the way of underlying regulations that change anything.

27:00Unidentified speaker (UNKNOWN): And

27:01Damian Classen: what do I mean by that? Well, you know, you see these regulations about um, oh, we're not going to let data centers go ahead um you know unless they're recycling their water, or we're not going gonna to let them use use our water or whatever the whatever the campaign the slogan. is. Well yeah it costs millions of dollars for them to put in systems to to recycle the water but relative to the building of a data center it's peanuts and would they do it if you didn't make them? No they wouldn't recycle the water if you didn't make them do it. But if you put the regulation in to make them do it, well they're just gonna do it. Like it's not a it's not it's nothing really gonna that's really gonna stop the data center. Um The other thing we s I see quite often is these statements about we're just not gonna let we're not gonna let these data centers connect to the grid. And these data centers are no longer connecting to the grid anyway. They couldn't get there, well they're not we're not gonna let them connect to the grid for the next three years or something like that tends to be the the statement that the politician makes. And the thing is these guys couldn't connect to the grid. There's no grid space available. So they're already putting in all their own power. Yes they've put in things to connect to the grid but they're trying to get it connected to the grid in four years or five years or whatever it is. And so it's effectively a meaningless political statement. And if that continues, then I don't I don't see any real change to the AI CapEx. But if that does slow down, um, then that's an issue for AI CapEx.

28:15Unidentified speaker (UNKNOWN): Uh and

28:16Damian Classen: I'll I'll talk a little bit more about what that actually means in a second, because that's so that's our first part is the As political forces and AI CapEx. The second part is the capital markets and AI CapEx. So there's sort of three big things that I think about in terms of this. And so what this means is that look, um, we need money, you know, the capital markets, so we need money actually to build these data centers. And where's that coming from? There's sort of three big places it comes from. One is um debt markets to the hyperscalers, so um Google,

28:47Unidentified speaker (UNKNOWN): um

28:48Damian Classen: Amazon, Microsoft. And at the moment, they are completely open. Look, the rates they're paying have gone up and then gone up a lot, I guess, relative to government bonds. But these things all had net cash. And so you sort of basically gone from, okay, they've they've gone from being priced almost like a government to now they're being priced more like a company. There hasn't been a significant change. If Google wants to lend money, there is no problems with people lending money to Google. Same with Microsoft. They don't have that much debt. Those debt markets remain open. Now, if they close, that's a big issue. But at the moment, though, that's your biggest part. That's pretty much open. The next part is equity markets to places like Anthropic, SpaceX, OpenAI are your big ones. But also they meta did a capital raising recently and things like that. So that money is being taken, you yeah, know, raised capital, put back into um building the these data centers. That's a big issue. They're the they're your real big ones, your three big ones anthropic, openai, uh SpaceX, these sort of um your major ones on the um uh sort of loss making company fronts. A lot of the neo clouds that are popping up, look, they're bit players. Look, yeah, that

29:55neo cloud didn't build that data center. Somebody else would have built the data center, is is my rough view on this. Yeah, if if um core. Weave can't get its funding away because it's just geared itself far too much, um, yeah, then the data center they want it to build will get built by Microsoft or OpenAI or something like that instead. So those equity markets, yeah. So the sort of debt markets to near clouds, to me, is a little bit of a side issue.

30:20Unidentified speaker (UNKNOWN): Um uh

30:21Damian Classen: it is indicative about how fast the boom will grow and things like that. But if if say debt markets are closed. to Neo Clouds, but the other debt markets to hyperscalers and equity markets to these other companies stays open, the AI CapEx boom is going ahead. So there's really not a lot they can do about that. So we've got this uh CapEx boom. That's that's our first part. Now, within the CapEx, the slower that goes, the worse it is for NVIDIA. Um, but it's probably a good thing for the hyperscalers. hyperscales. So basically, if if nobody can build, or nobody can build as fast as they want because of either shortages, uh political, um, you know, uh political bans in terms of stopping these data centers from going as fast.

31:00Unidentified speaker (UNKNOWN): Um

31:01Damian Classen: and uh so what's gonna happen then is as long as you know, and we haven't talked about demand for AI yet, we're gonna talk about that second, but as long as that demand from AI AO is ticking along, then all it means is it's a negative for people who are selling stuff into it. Nvidia, Broadcom, any of these guys that actually make it putting um chips into it because they can't sell as many as they want. But the hyperscalers, it's a good thing.

31:24Unidentified speaker (UNKNOWN): They're

31:24Damian Classen: their NVIDIA chip costs go down. Um, you know, they've got they they have the um the capacity, there's demand for the capacity is shooting up and and the supply isn't coming on as fast. And so they get to charge overs oversized prices, slowing CapEx is. certainly a bad thing for NVIDIA and bad thing for a lot of those companies, but it's a good thing for other players within the sector. So

31:46Unidentified speaker (UNKNOWN): I

31:47Damian Classen: guess where I'm looking at from this is we have this CapEx issue. We certainly need to be watching those political, we need to be watching capital markets, but um it's not a sign for the whole AI boom. It's it's a sign about which which players within the AI boom you want to take advantage of.

32:02Unidentified speaker (UNKNOWN): Okay,

32:02Damian Classen: that's one part of it. Second part then is the AI demand for AI. That's that's the much more important thing for the for the AI boom. And we really don't get much from uh about what's happening with NVIDIA from that. But it's what really matters in the end for everything. And again, I'd split this into two parts. One is the demand for AI training. So what I'm talking about here is

32:26Unidentified speaker (UNKNOWN): we

32:26Damian Classen: have Anthropic, OpenAI, Google, Meta, a whole bunch of these companies, SpaceX, all these companies out there with AI models that need to be trained, a whole bunch of Chinese ones as well. It's an arms race within that.

32:41Unidentified speaker (UNKNOWN): If

32:41Damian Classen: you are the if if Anthropic was the only one with it or OpenAI was the only one with it, and there was nobody else, then they could just keep going, yeah, we'll spend whatever on RD and you know, we'll we'll keep it. spend enough to keep this thing moving forward, but we don't need to overspend on it because uh you know there's only one or two of us out there, and you know, we can keep that keep that going. At the moment, that's not the case. At the moment, this is everyone's out there going, Well, it's early days. If I can get a better model than everyone else, then I'll be in the front. And um, so you know, it's it's a rush. Everyone's trying to spend that money. The amounts are getting bigger and bigger. and so people, the smaller players are dropping off down the bottom, but there's there's a fee for being the having the frontier model and the best model out there that you can charge. And it's probably a bit of winners take all or close to. There's probably only gonna be a couple winners in this. Uh and so the impetus for all these companies is to keep spending money. And and it's and it is an arms race. And so uh that arms race though. is is is funded by capital markets. So if capital markets remain open dot anthropic, which is listing tomorrow, uh OpenAI, uh and SpaceX, then that sort of underpins a lot of demand from this. And so it's not the most important thing, but it is certainly uh gonna keep that that sort of dollars per GPU hour nice and high. If you can sort of go, well, I can

34:01I've just got my I've got my chips, I want so to rent them out to people to to to do things. Um if I can underpin 20% or 30% or 40% of my revenues from from these models that want to run, then great. That that's a big company's willing to spend lots of money. So it's important, but uh, but I guess again, this is something where if

34:21Unidentified speaker (UNKNOWN): if

34:21Damian Classen: some of these companies start falling by the wayside, um, but the the inference part, people actually using AI stuff. stay strong, then I I think you can get through a capital cycle where we're spending the growth in in AI training um is not significant or and maybe even sideways or or even a little bit backwards. I think the the the AI boom could stand up to that if uh the last part, the AI inference, people actually using AI is the stuff that that ticks on. So uh yeah. it's good good to keep things going fast if you if those capital markets are open, but not as sort of necessary to keep the whole thing going. And so then that leads to that fourth point, which is the key one. It's the demand for AI inference and the amount of money people are willing to spend on that. And I think in the end, uh token prices. I've said this plenty of times, but I'll keep reiter reiterating reiterating it. Don't look at token prices. The number of tokens. being produced is increasing exponentially, and the token prices are falling exponentially. And so trying to work out where that's going is really hard. Look at GPU cost per hour. That's that's the key to this because in the end,

35:28Unidentified speaker (UNKNOWN): um,

35:29Damian Classen: that's really gonna drive that's that's showing the demand that people are using. Because if token prices um decrease by 90%, but I add a zero to the number of tokens that are. I use, then net effect is

35:41Unidentified speaker (UNKNOWN): um

35:41Damian Classen: you know that still that that dollars is still being spent out out there. Is you're ending up with a I'm spending the same amount for and just getting more for it. So uh yeah, so within that, it's really about this demand from the rest of the economy. And if we start seeing real weakness from that, then uh that's the real key to the end of the cycle and things finishing. And where do I see this differently say to the um uh the tech boom is that I think within the tech boom, a large part of what was happening with it was at the peaks of the boom, not in the early stages of the boom. The early stages of the boom was, I think, much more sustainable in terms of the spending and what was happening. But as the boom was peaking and sort of going too far, so to speak, we ended up up in this scenario where companies were raising money. The um, you know, pets.com as you know, the the uh um

36:31Unidentified speaker (UNKNOWN): you know,

36:31Damian Classen: the the poster child for the boom. Companies were raising money to do stuff on the internet, they didn't quite know what they were going to do. They'd raise this money and then then they'd take that money and they'd turn around and they'd spend it on um uh basically other companies within the industry. They'd basically go, Let's spend on advertising on other companies uh in order to get more eyeballs, and then eventually we'll we'll be. the winners. And so there was definitely a large part of that was people were raising money to then give to other companies that was in their revenue, which they'd give to other companies. And that sort of kept the whole thing alive or kept some revenue coming in while they were they were mostly loss making. This is different. And the this is more about the you Yeah, know, the the money that's being raised now is being put into data centers to create uh the capacity for. people to be able to type in queries and and get an answer back. And the and the queries are getting better and better, exponentially better over time. So

37:23Unidentified speaker (UNKNOWN): it

37:23Damian Classen: really comes back to that end demand. If end demand from the rest of the economy keeps growing and there's I can't see any signs really of that slowing down, then it's pretty hard to say that there's an end of the cycle at the moment. And the other part is when you're talking about the

37:39Unidentified speaker (UNKNOWN): the

37:39Damian Classen: NVIDIA, which is sort of you know right at the the the forefront uh of the AI boom being priced on 10 times earnings or seven times the 10 times earnings for next year and seven times earnings for the year after, it's pretty hard to argue that's a massive bubble that that nobody heard of. You know, when you're talking about these things being priced on a hundred times earnings, uh, you know, and that was where we were in the in the a in the tech boom was companies like um uh Cisco being priced on a hundred times forward earnings which you know kept admittedly you know were overpriced earnings as well and so earnings fell a bit and you saw the the the the multiples come back to more reasonable revel levels. That's where you see a big boom that sort of ends up in a bust. Uh the issue about having companies that are much lower and much more reasonable multiples uh I think is um yeah it's it's pretty hard to argue that there's a you know there's it's it's hard to I don't think you can argue there's a pricing boom at the moment. You can argue there's a profitability boom that that's higher than what it should be and that profits will disappear. And I think there's potentially a case in some of these. And and I think Nvidia is being priced already that that its earnings will be lower. Um, And so I think there's uh you know that's certainly an issue that that we need to consider. But the um we're certainly not looking at a an sort of an an earnings type boom at the moment, which we need to worry about that. Um sorry, we're not looking at a pricing boom,

38:59which you need to worry about unraveling. Now it's just a question about the earnings side. So while that's the case, um, you know, for for me it's all about coming back and focusing really on those earnings. Where are we starting to see any weakness in that? Um,

39:12Unidentified speaker (UNKNOWN): you

39:12Damian Classen: know, certainly nothing from NVIDIA at the moment. but I think it's worth keeping them in mind in terms of that whole picture of the the industry and and what's happening and just understanding that um, you know, the neo clouds, for example, are having some debt problems. Um, they're a very small part of the industry and they're the highly leveraged part, and it was driven by some excesses. Uh, I think the the industry could certainly survive without them. Uh, and if you don't have them, it's it's good for um hyperscalers, probably bad for Nvidia's and Broadcoms and companies like that, but um the net effect for the whole cycle is is is not like an end-to-cycle event. Anyway, that was uh all I sort of wanted to talk about on uh NVIDIA today. Uh I'll be writing up a bit more on that and um I'm sure we'll be uh talking more in future weeks. Uh thanks to for tuning into Nucleus Investment Insights. If you found today's insights helpful, give us a like and subscribe so you don't miss any future episodes. If you've got any questions, leave them in the comments. We'll answer them on our socials. Don't forget to catch us live for the next episode, Thursday, 12:30 Australian to Eastern Australian time. Uh, for more insights, visit our website, sign up for our weekly email to get updates straight in your inbox. And if you want to build a personalized investment portfolio, you can head to our website, start your risk-free account, send us an email or book a free consultation. tool with our team.

This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights