Breaking News To Trading Moves

Cerebras raises 2026 revenue forecast amid AI chip diversification

16 min Listen to the original Episode page
AI HardwareSemiconductorsCloud InfrastructureMarket DiversificationSupply ChainRevenue ForecastsTech StocksInvestment Risks

Summary

Speakers discuss Cerebras Systems raising its 2026 core revenue forecast to $940–$960 million, driven by new customer Microsoft and broader AI hardware demand. The episode analyzes whether this signals a market shift away from AI hardware monopolies toward alternative architectures to avoid vendor lock-in. Speakers identify winners including Cerebras, AMD, Microsoft, Oracle, TSM, and AMAT, while noting risks for established competitors like Nvidia and infrastructure operators facing asset depreciation. Execution risks and supply chain dependencies are highlighted as critical factors for these companies.

Key takeaways

  1. 1
    Cerepras also announced Microsoft as a new customer, which visibly strengthens their position in AI computing infrastructure.

    Unidentified speaker (speaker_0) · 1:19

  2. 2
    They benefit directly from integrating additional computing architectures that reduce AI processing costs or improve performance.

    Unidentified speaker (speaker_2) · 5:45

  3. 3
    applied materials benefits if this sustained demand encourages manufacturers to invest in additional capacity.

    Unidentified speaker (speaker_2) · 8:21

Episode sections

  1. 0:00 – 1:51

    Introduction and Cerebras Revenue Update

    Speakers introduce the topic of rapid AI hardware obsolescence and detail Cerebras Systems raising its 2026 core revenue forecast to $940–$960 million, up from $880–$890 million, with Q3 revenue of $230.9 million. The announcement includes Microsoft as a new customer, driving a >17% after-hours stock surge.

  2. 1:51 – 2:43

    Market Shift Toward Alternative AI Chips

    The discussion analyzes whether the forecast indicates a market shift away from AI hardware monopolies, concluding that buyers are actively diversifying budgets toward alternative architectures to avoid vendor lock-in and supply chain bottlenecks.

  3. 2:43 – 4:21

    Winners: Alternative Chip Manufacturers

    Cerebras Systems ($CBRS) and Advanced Micro Devices ($AMD) are identified as winners due to increased demand for alternative computing architectures, with AMD benefiting as cloud providers diversify beyond Nvidia.

  4. 4:21 – 5:30

    Execution Risks and Supply Chain Dependencies

    Speakers clarify that demand does not guarantee market share gains; Cerebras and AMD must execute on technology, software integration, and physical supply chain delivery, with Cerebras relying on foundries like TSMC for manufacturing.

  5. 5:30 – 7:37

    Winners: Cloud and Infrastructure Providers

    Microsoft ($MSFT) benefits directly as a Cerebras customer through improved performance or lower costs, while Oracle ($ORCL) benefits indirectly from broader enterprise AI inference demand, even without a direct Cerebras link.

  6. 7:37 – 9:26

    Winners: Semiconductor Manufacturing and Equipment

    Taiwan Semiconductor Manufacturing ($TSM) and Applied Materials ($AMAT) are highlighted as beneficiaries of increased manufacturing orders and capital expenditure cycles, driven by the need to expand production capacity for alternative AI chips.

  7. 9:26 – 12:21

    Losers: Established Competitors and Traditional Processors

    Nvidia ($NVDA) and Broadcom ($AVGO) face pressure from budget reallocation to alternatives, while Intel ($INTC) and AMD face headwinds in traditional server processors as data centers prioritize specialized AI hardware.

  8. 12:21 – 15:37

    Losers: Infrastructure Operators and Depreciation Risks

    CoreWeave ($CRWV) and Nebius ($NBIS) face risks from asset depreciation and financing costs if newer, cheaper hardware renders their existing leased or owned infrastructure less competitive before investment recovery.

Assets discussed

AssetKindStanceContext
MicrosoftMSFTequitybullishCerepras also announced Microsoft as a new customer, which visibly strengthens their position in AI computing infrastructure.
NVIDIANVDAequitybearishWhich brings us to the established losers. First, the established AI competitors, NVIDIA. ticker NVDA, and Broadcom, ticker AVGO. Aaron

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

Claims worth verifying

Cerebras Systems raised its 2026 core revenue forecast to a range of $940 million to $960 million.

Unidentified speaker (UNKNOWN) ·0:45

How to verify: Check Cerebras Systems Q3 2026 earnings release or press statement on their investor relations page or SEC EDGAR (if public) or official corporate communications. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Primary source

The previous outlook for Cerebras 2026 core revenue was $880 million to $890 million.

Unidentified speaker (speaker_2) ·0:57

How to verify: Check prior Cerebras Systems guidance in previous earnings releases or press statements. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Third-quarter core revenue for Cerebras was $230.9 million.

Unidentified speaker (speaker_0) ·1:10

How to verify: Check Cerebras Systems Q3 2026 earnings report or financial statements. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Third-quarter core revenue represented approximately 70% year-over-year growth.

Unidentified speaker (speaker_0) ·1:10

How to verify: Compare Q3 2026 core revenue ($230.9M) with Q3 2026 core revenue from Cerebras financial reports. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Microsoft was announced as a new customer for Cerebras.

Unidentified speaker (speaker_1) ·1:19

How to verify: Check Cerebras Systems press release or Microsoft's supplier announcements/newsroom. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Primary source

Cerebras shares surged more than 17% in after-hours trading following the announcement.

Unidentified speaker (UNKNOWN) ·1:28

How to verify: Check stock price data for $CBRS on financial platforms (e.g., Bloomberg, Yahoo Finance) for the trading day following the announcement. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Cerebras does not have its own factories and relies on foundries for manufacturing.

Unidentified speaker (speaker_2) ·5:21

How to verify: Check Cerebras Systems business model description in their investor presentations or SEC filings (if public) or company website. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Primary source

CoreWeave ticker is CRWV and Nebius ticker is NBIS.

Unidentified speaker (speaker_0) ·12:47

How to verify: Check stock ticker symbols on financial data providers (e.g., Bloomberg, Reuters) for CoreWeave and Nebius. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Primary source

Nvidia dominates the AI chip space but faces competition from alternative architectures.

Unidentified speaker (UNKNOWN) ·9:42

How to verify: Check market share reports from firms like IDC or Gartner on AI accelerators, and Nvidia's market position statements. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Primary source

Intel faces challenges competing in advanced computing markets as spending priorities shift to specialized AI hardware.

Unidentified speaker (UNKNOWN) ·11:11

How to verify: Check Intel's financial reports and industry analysis on data center spending trends shifting from CPUs to AI accelerators. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify Cerebras Systems' official 2026 revenue guidance and Q3 financial results via their investor relations page or SEC filings.
  • Check Microsoft's public statements or earnings calls regarding their specific AI infrastructure spending and Cerebras partnership details.
  • Review AMD's recent earnings reports and guidance for cloud provider diversification trends beyond Nvidia.
  • Analyze TSMC and Applied Materials' capital expenditure plans and capacity expansion announcements for semiconductor manufacturing.
  • Assess Nvidia and Broadcom's recent market share data and competitive positioning against alternative AI chip architectures.
  • Evaluate CoreWeave and Nebius' asset depreciation schedules and financing costs in their respective financial filings.
  • Research the technical specifications and performance metrics of Cerebras' WaferScale Engine compared to Nvidia's GPUs.
  • Look for counter-arguments or analyst reports questioning the sustainability of AI hardware demand or the viability of alternative architectures.
Full transcriptexpand
0:00
Unidentified speaker (speaker_0): Welcome to Breaking News to Trading Moves. Imagine spending $50 million on a supercomputer today, only to find out next year that a cheaper, faster chip just made your, you know, your massive physical investment totally obsolete.
0:14
Unidentified speaker (speaker_1): Yeah, that is a terrifying prospect for any operator holding those assets.
0:17
Unidentified speaker (speaker_0): It really is. And that is the exact scenario playing out in the artificial intelligence computing sector right
0:23
Unidentified speaker (UNKNOWN): now.
0:23
Unidentified speaker (speaker_0): We have a stack of articles and financial notes detailing a headline. that is forcing the market to reevaluate who really holds the power.
0:31
Unidentified speaker (speaker_1): Right.
0:31
Unidentified speaker (speaker_0): AI chipmaker Cerebrus Systems has officially raised its annual revenue forecast, and they are pointing directly to accelerating demand for AI computing infrastructure.
0:41
Unidentified speaker (speaker_2): And the numbers here give us a very precise picture of their current trajectory.
0:45
Unidentified speaker (UNKNOWN): I mean,
0:45
Unidentified speaker (speaker_2): Cerebrus raised its 2026 core revenue forecast to a range of 940 million. to $960 million.
0:53
Unidentified speaker (speaker_0): Okay. And just for context for you listening, what was the previous outlook?
0:57
Unidentified speaker (speaker_2): Right. So the previous outlook was sitting at $880 million to $890 million. So it is a clear bump up. And when you look at their most recent quarter, well, third quarter core revenue hit $230.9 million.
1:10
Unidentified speaker (speaker_0): Whoa.
1:10
Unidentified speaker (speaker_2): Yeah. Which represents approximately 70% year over year growth.
1:14
Unidentified speaker (UNKNOWN): That
1:14
Unidentified speaker (speaker_0): is massive. growth. And there is another major catalyst attached to this headline, too.
1:19
Unidentified speaker (speaker_1): Yes, there is.
1:19
Unidentified speaker (speaker_0): Cerepras also announced Microsoft as a new customer, which visibly strengthens their position in AI computing infrastructure.
1:27
Unidentified speaker (speaker_2): Oh, absolutely.
1:28
Unidentified speaker (UNKNOWN): The
1:28
Unidentified speaker (speaker_0): market reacted immediately to that, with shares surging more than 17% in after hours trading. So for you listening, our mission today is straightforward.
1:37
Unidentified speaker (UNKNOWN): We are
1:38
Unidentified speaker (speaker_0): reviewing this headline. and laying out the specific winners and losers in the market based strictly on this announcement.
1:44
Unidentified speaker (speaker_2): And the data we have. Yeah.
1:45
Unidentified speaker (speaker_0): Exactly. But before we get to the specific companies, I have a broader setting question for you.
1:50
Unidentified speaker (speaker_2): Sure. Go ahead.
1:51
Unidentified speaker (speaker_0): Does this forecast change indicate that the broader market is actively seeking alternatives to um current AI hardware monopolies? Aaron
1:59
Unidentified speaker (UNKNOWN): Powell
1:59
Unidentified speaker (speaker_2): The data strongly suggests exactly that. I mean when a challenge Challenger company in the semiconductor space raises revenue forecasts by this margin and you know secures a primary cloud provider like Microsoft.
2:11
Unidentified speaker (speaker_0): That is a big deal.
2:12
Unidentified speaker (speaker_2): It is. It shows the buyers are actively allocating budgets toward alternative architectures. Relying on a single dominant supplier just creates bottlenecks.
2:21
Unidentified speaker (speaker_0): Right, because buyers are constantly evaluating cost and performance.
2:24
Unidentified speaker (speaker_2): Cost, performance, and supply chain availability. They want options.
2:28
Unidentified speaker (UNKNOWN): And
2:28
Unidentified speaker (speaker_2): this forecaster adjustment tells us they are willing to put actual capital behind those options to ensure they don't get locked into one vendor's pricing or uh their production timeline.
2:38
Unidentified speaker (speaker_0): Makes sense. So we should logically start by looking at the direct beneficiaries here, right?
2:43
Unidentified speaker (speaker_2): Yeah, the alternative AI chip manufacturers.
2:45
Unidentified speaker (speaker_0): Right. First on the winner's list we have Cerebrus Systems, ticker C B CBRS, R.
2:51
Unidentified speaker (speaker_0): and Advanced Microdevices, ticker AMD.
2:54
Unidentified speaker (UNKNOWN): So
2:55
Unidentified speaker (speaker_2): Cerebrus benefits directly from its own stronger revenue expectations and its newly expanding customer bases we just established.
3:01
Unidentified speaker (speaker_0): Obviously, yeah.
3:02
Unidentified speaker (UNKNOWN): And
3:02
Unidentified speaker (speaker_2): then you look at advanced microdevices or AMD, they stand to benefit as cloud providers diversify their AI hardware purchases beyond NVIDIA. Aaron
3:10
Unidentified speaker (UNKNOWN): Powell
3:10
Unidentified speaker (speaker_0): Right. They are the next logical option. Aaron
3:12
Unidentified speaker (UNKNOWN): Powell
3:13
Unidentified speaker (speaker_2): Exactly. Both of these companies. offer exposure to the demand for alternative computing architectures. When cloud providers decide they need to spread their bets, companies designing these alternative systems are the immediate landing spot for those hardware budgets. Aaron
3:27
Unidentified speaker (UNKNOWN): Powell
3:27
Unidentified speaker (speaker_0): Is this um
3:28
Unidentified speaker (UNKNOWN): is
3:29
Unidentified speaker (speaker_0): this like airlines finally buying planes from Airbus instead of just Boeing so they can diversify their fleet? Aaron
3:33
Unidentified speaker (UNKNOWN): Powell
3:34
Unidentified speaker (speaker_2): That is a great way to think about it. Aaron
3:35
Unidentified speaker (UNKNOWN): Powell
3:35
Unidentified speaker (speaker_0): Because you don't want your entire operation grounded if one manufacturer has a supply issue, right? And you definitely want leverage when negotiating prices. Aaron
3:44
Unidentified speaker (UNKNOWN): Powell
3:44
Unidentified speaker (speaker_2): That analogy works, but let's push it one step further.
3:47
Unidentified speaker (speaker_3): Okay.
3:48
Unidentified speaker (speaker_2): Airlines don't just buy different brands of planes, they buy different types of planes for different routes.
3:54
Unidentified speaker (speaker_2): Like uh massive jumbo jet handles international flights. Aaron
3:57
Unidentified speaker (UNKNOWN): Powell
3:57
Unidentified speaker (speaker_3): Exactly. While a smaller plane handles short regional hops. Aaron
4:00
Unidentified speaker (UNKNOWN): Ross Powell
4:01
Unidentified speaker (speaker_2): And the same logic applies to AI workers. workloads. Cloud providers need different computing architectures because some chips are incredibly efficient at, say, training AI models.
4:10
Unidentified speaker (speaker_0): Well, others might be better or cheaper at simply running them.
4:13
Unidentified speaker (speaker_2): Exactly. Fostering a competitive supplier ecosystem gives them the exact tools they need for specific tasks.
4:21
Unidentified speaker (UNKNOWN): Aaron Powell
4:21
Unidentified speaker (speaker_0): But let me push back a bit here. Does stronger industry demand guarantee that these specific alternative companies will actually gain market share? Because just wanting alternatives doesn't mean these specific chips from Cerebrus or AMD will, you know, perform the way the buyers actually need them to. Aaron
4:37
Unidentified speaker (UNKNOWN): Powell
4:37
Unidentified speaker (speaker_2): That is a crucial distinction. According to the data, stronger demand offers exposure, but it absolutely does not guarantee market share gains.
4:46
Unidentified speaker (speaker_0): Right.
4:46
Unidentified speaker (speaker_2): Cerebrus and AMD have the opportunity to capture that demand, but they still have to execute on the technology. Trevor
4:52
Unidentified speaker (UNKNOWN): Burrus
4:53
Unidentified speaker (speaker_0): They have to deliver the raw computing performance. Trevor Burrus
4:55
Unidentified speaker (speaker_2): Right. And they have to ensure the software integrates seamlessly for the end user. Plus, they have to manage the physical supply chain to actually deliver the chips.
5:02
Unidentified speaker (UNKNOWN): Aaron Ross Powell
5:03
Unidentified speaker (speaker_0): So the demand creates an open door, but it does not automatically translate to a permanent gain in market share. Trevor
5:08
Unidentified speaker (UNKNOWN): Burrus
5:09
Unidentified speaker (speaker_2): Not if the physical execution falls short, no.
5:11
Unidentified speaker (UNKNOWN): Okay,
5:11
Unidentified speaker (speaker_0): so if Cerebrus and AMD are successfully pulling in these mass massive budgets, where are they physically getting these chips made?
5:18
Unidentified speaker (speaker_2): Good question. Because
5:19
Unidentified speaker (speaker_0): Cerebrus doesn't have its own factories.
5:21
Unidentified speaker (speaker_2): No, they don't.
5:22
Unidentified speaker (speaker_0): So that takes us from the companies designing these alternative architectures to the cloud providers purchasing them
5:28
Unidentified speaker (UNKNOWN): and
5:28
Unidentified speaker (speaker_0): the foundries manufacturing them.
5:30
Unidentified speaker (UNKNOWN): Our
5:30
Unidentified speaker (speaker_0): next category of winners covers the cloud and infrastructure providers.
5:35
Unidentified speaker (UNKNOWN): We
5:35
Unidentified speaker (speaker_0): are looking at Microsoft, Ticker MSFT, and Oracle. ticker ORCL. Aaron
5:40
Unidentified speaker (UNKNOWN): Powell
5:41
Unidentified speaker (speaker_2): Starting with Microsoft, they are now an announced customer of Cerebrus.
5:45
Unidentified speaker (speaker_0): Right.
5:45
Unidentified speaker (speaker_2): They benefit directly from integrating additional computing architectures that reduce AI processing costs or improve performance. When Microsoft can process AI workloads more efficiently, they can offer more competitive pricing to their Azure Cloud customers.
6:01
Unidentified speaker (speaker_0): Or they can simply improve their own. profit margins. Trevor
6:03
Unidentified speaker (UNKNOWN): Burrus
6:03
Unidentified speaker (speaker_2): Exactly. Then we look at Oracle. They benefit from broader enterprise demand for AI infrastructure. As the overall appetite for AI processing grows across the corporate world, Oracle captures a portion of that enterprise spending. Aaron
6:17
Unidentified speaker (UNKNOWN): Powell
6:17
Unidentified speaker (speaker_0): Wait, I have to ask.
6:21
Unidentified speaker (speaker_0): But nothing in this Cerebrus announcement actually mentions them. Aaron
6:24
Unidentified speaker (UNKNOWN): Powell
6:24
Unidentified speaker (speaker_2): That's true.
6:24
Unidentified speaker (speaker_0): So how do they get dragged into this conversation if they aren't directly linked to Cerebrus in this specific headline? Connect the dots for me there.
6:33
Unidentified speaker (UNKNOWN): Aaron Powell
6:33
Unidentified speaker (speaker_2): Yeah, it comes down to the broader mechanics of how the cloud market functions, specifically regarding something called AI inference.
6:41
Unidentified speaker (speaker_0): Okay.
6:42
Unidentified speaker (speaker_2): To put it simply, training in AI is when it reads the internet to get gets smart. Aaron Powell
6:46
Unidentified speaker (speaker_0): Right, which takes massive concentrated supercomputers. Aaron
6:49
Unidentified speaker (UNKNOWN): Powell
6:49
Unidentified speaker (speaker_2): Right. Inference, on the other hand, is when you actually ask the AI a question and it generates an answer.
6:54
Unidentified speaker (speaker_0): Oh, okay.
6:55
Unidentified speaker (speaker_2): And inference happens millions of times a
6:58
Unidentified speaker (UNKNOWN): day.
6:58
Unidentified speaker (speaker_2): Growing AI inference demand broadly supports cloud providers that offer scalable computing capacity. Aaron
7:04
Unidentified speaker (UNKNOWN): Powell
7:04
Unidentified speaker (speaker_0): Making it an indirect opportunity for Oracle.
7:07
Unidentified speaker (speaker_2): Exactly.
7:08
Unidentified speaker (UNKNOWN): Even
7:08
Unidentified speaker (speaker_2): if Oracle isn't buying from Cerebrus today,
7:11
Unidentified speaker (UNKNOWN): the
7:11
Unidentified speaker (speaker_2): fact that AI infrastructure demand is accelerating, which Cerebrus forecast proves, means there is a higher total volume of AI workloads needing a home.
7:20
Unidentified speaker (speaker_0): Ah, I see. The rising tide of enterprise AI adoption lifts the platforms that host those daily applications.
7:26
Unidentified speaker (speaker_2): Precisely.
7:27
Unidentified speaker (speaker_0): That makes a lot of sense.
7:28
Unidentified speaker (UNKNOWN): The
7:29
Unidentified speaker (speaker_0): volume of daily usage just keeps going up. But we also have to factor in the physical hardware side. Designing a processor is one thing.
7:37
Unidentified speaker (speaker_2): Actually, fabricating it from raw silicon is entirely different.
7:41
Unidentified speaker (speaker_0): Exactly. That brings us to the manufacturing and equipment winners. Taiwan Semiconductor Manufacturing, Ticker T S M, and Applied Materials, Ticker AMAT.
7:50
Unidentified speaker (speaker_2): Right. So TSM C is the foundry. They physically manufacture the chips for many of these designers, including the alternatives.
7:57
Unidentified speaker (UNKNOWN): So
7:57
Unidentified speaker (speaker_0): they benefit from increased demand for advanced semiconductor manufacturing as AI chip developers expand production. Aaron
8:03
Unidentified speaker (UNKNOWN): Powell
8:03
Unidentified speaker (speaker_2): Yes. If Cerebrus is projecting higher revenue, it means they need to produce and sell more physical units, which means more manufacturing orders for TSMC.
8:11
Unidentified speaker (speaker_3): Got it.
8:12
Unidentified speaker (speaker_2): Then you go one step further up the supply chain to applied materials. They manufacture the highly specialized equipment that TSMC uses to build the chips in the first place.
8:21
Unidentified speaker (UNKNOWN): So
8:21
Unidentified speaker (speaker_2): applied materials benefits if this sustained demand encourages manufacturers to invest in additional capacity.
8:26
Unidentified speaker (speaker_3): I want to focus on applied materials for a second because the mechanics there are fascinating. It sounds like a chain reaction of capital expenditure. Aaron
8:34
Unidentified speaker (UNKNOWN): Powell
8:34
Unidentified speaker (speaker_2): Yes, it really is. The semiconductor supply chain operates on massive capital expenditure cycles,
8:40
Unidentified speaker (UNKNOWN): and
8:40
Unidentified speaker (speaker_2): it's prone to something called the bull whip effect. Aaron
8:43
Unidentified speaker (UNKNOWN): Ross Powell
8:43
Unidentified speaker (speaker_0): Okay. How does that work? Aaron
8:44
Unidentified speaker (UNKNOWN): Ross Powell Well
8:44
Unidentified speaker (speaker_2): Cerebrus Reverse gets a new order from Microsoft. Cerebers
8:46
Unidentified speaker (UNKNOWN): So
8:47
Unidentified speaker (speaker_2): Reverse then pays TSMC to make the chip.
8:49
Unidentified speaker (UNKNOWN): But
8:50
Unidentified speaker (speaker_2): if TSMC is already running in their factories at maximum capacity, they can't just flip a switch to make more.
8:56
Unidentified speaker (speaker_0): They have to expand.
8:57
Unidentified speaker (speaker_2): Right. They have to build entirely new factory lines.
9:00
Unidentified speaker (UNKNOWN): That
9:00
Unidentified speaker (speaker_2): means TSMC pays applied materials for the machinery required to build those lines.
9:04
Unidentified speaker (speaker_0): And building those machines takes time, right?
9:06
Unidentified speaker (speaker_2): It takes months, sometimes years.
9:08
Unidentified speaker (UNKNOWN): And
9:08
Unidentified speaker (speaker_2): it requires enormous capital. These are indirect opportunities, sure, but they are vital to the ecosystem.
9:15
Unidentified speaker (UNKNOWN): So
9:15
Unidentified speaker (speaker_0): if cloud providers and enterprises are directing billions of dollars toward alternative architectures and specialized AI accelerators, those dollars have to be reallocated from existing budgets.
9:25
Unidentified speaker (speaker_2): Yes, they do.
9:26
Unidentified speaker (speaker_0): It is a zero-sum game
9:27
Unidentified speaker (UNKNOWN): that
9:28
Unidentified speaker (speaker_0): puts a target on the back of the established competitors and traditional processor suppliers.
9:32
Unidentified speaker (speaker_2): It absolutely does. Aaron
9:33
Unidentified speaker (UNKNOWN): Powell
9:33
Unidentified speaker (speaker_0): Which brings us to the established losers. First, the established AI competitors, NVIDIA. ticker NVDA, and Broadcom, ticker AVGO. Aaron
9:42
Unidentified speaker (UNKNOWN): Powell
9:42
Unidentified speaker (speaker_2): Right. So NVIDIA has dominated this space. But they now face the reality that customers may allocate future spending to alternative architectures.
9:51
Unidentified speaker (speaker_0): Because enterprise
9:51
Unidentified speaker (speaker_3): budgets are finite.
9:53
Unidentified speaker (speaker_2): Exactly. And data centers also have strict physical limits.
9:57
Unidentified speaker (UNKNOWN): A
9:57
Unidentified speaker (speaker_2): building only has so much floor space. And more importantly, it only has so much electricity coming from the grid. Aaron
10:02
Unidentified speaker (UNKNOWN): Powell
10:03
Unidentified speaker (speaker_0): So if a cloud provider allocates power and budget to a new Cerebrus or AMD cluster. Aaron
10:08
Unidentified speaker (UNKNOWN): Powell
10:08
Unidentified speaker (speaker_2): That is power and budget not going to NVIDIA.
10:11
Unidentified speaker (speaker_0): Wow, right. Aaron
10:11
Unidentified speaker (UNKNOWN): Ross Powell And
10:12
Unidentified speaker (speaker_2): Broadcom also faces competition in selected accelerator opportunities. However, it is vital to know that Broadcom's custom silicon and networking businesses still benefit from overall AI investment. Aaron
10:23
Unidentified speaker (UNKNOWN): Powell
10:23
Unidentified speaker (speaker_0): So they aren't losing out entirely. Aaron
10:25
Unidentified speaker (UNKNOWN): Powell
10:25
Unidentified speaker (speaker_2): No. They are a loser in this. the specific context of alternative accelerators taking some of the spotlight, but their broader business model has insulation. Aaron
10:33
Unidentified speaker (UNKNOWN): Powell
10:33
Unidentified speaker (speaker_0): Okay, that makes sense. We also have the traditional processor suppliers on the losers list. We are looking at Intel, ticker INTC, and advanced microdevices, ticker AMD.
10:43
Unidentified speaker (speaker_2): Right. As AI With infrastructure budgets heavily favors specialized accelerators, conventional server processor supplies. suppliers face severe pressure on their product mix and pricing.
10:53
Unidentified speaker (UNKNOWN): Aaron Powell
10:54
Unidentified speaker (speaker_0): Because historically, data centers spent the bulk of their money on traditional central processing units or CPUs. Aaron
11:00
Unidentified speaker (UNKNOWN): Powell
11:00
Unidentified speaker (speaker_2): Yes, exactly. But now the priority is specialized AI hardware. If you are building a new data center today, you are dedicating the maximum possible footprint to AI chips.
11:11
Unidentified speaker (UNKNOWN): Aaron Powell
11:11
Unidentified speaker (speaker_0): Which leaves much less room for traditional servers.
11:15
Unidentified speaker (speaker_2): Precisely. And Intel faces distinct challenges competing across these advanced computing markets as that spending priority physically changes the layout of the modern data center.
11:26
Unidentified speaker (speaker_0): Wait, I spot a contradiction in our notes here. We just listed advanced microdevices, ticker AMD, as a winner in the alternative chip category. Now they are right here on the losers list. Can you reconcile how a single company can be both a winner and a loser in this exact scenario?
11:42
Unidentified speaker (speaker_2): Yeah, I know it sounds contradictory, but AMD has a highly mixed outlook because of the two different sides of their business.
11:49
Unidentified speaker (speaker_0): Okay, explain that.
11:50
Unidentified speaker (speaker_2): Their AI accelerators benefit from the diversification trend we discussed earlier that makes them a winner. However, they also have a massive legacy business selling traditional server processors.
12:01
Unidentified speaker (speaker_0): Ah, I see.
12:02
Unidentified speaker (speaker_2): And that side of the business is exposed to the exact same changing spending priorities that are pressuring Intel.
12:08
Unidentified speaker (UNKNOWN): So
12:08
Unidentified speaker (speaker_0): while their new AI chips are gaining traction and winning budgets, their traditional processor sales are facing declining budgets.
12:15
Unidentified speaker (speaker_1): Exactly.
12:16
Unidentified speaker (speaker_2): They win in one product category and face strong headwinds in another.
12:20
Unidentified speaker (speaker_0): Got
12:21
Unidentified speaker (UNKNOWN): it.
12:21
Unidentified speaker (speaker_0): So beyond the designers and the legacy chipmakers, we must look at the operators who purchase and hire
12:26
Unidentified speaker (UNKNOWN): the channel.
12:27
Unidentified speaker (speaker_0): host this hardware.
12:28
Unidentified speaker (speaker_2): Right.
12:28
Unidentified speaker (speaker_0): Because rapid improvements in alternative hardware carry distinct financial risks for those holding the physical assets. This brings us to the infrastructure risk losers, specifically operators facing cost pressure. We are looking at Core Weave, ticker CRWV, W V and Nebius, ticker
12:47
Unidentified speaker (speaker_0): NBIS. Aaron
12:48
Unidentified speaker (UNKNOWN): Powell
12:49
Unidentified speaker (speaker_2): Yeah, this is a matter of, well. asset depreciation. Rapid improvements in AI hardware create risks for companies investing heavily in computing infrastructure.
12:58
Unidentified speaker (speaker_0): Right.
12:59
Unidentified speaker (speaker_2): Operators like Core Weave and Nebius purchase massive amounts of hardware to rent out computing power to other businesses. They could face pressure if the expense of hardware they purchased becomes less competitive before they recover their initial investment.
13:13
Unidentified speaker (speaker_0): It sounds exactly like buying the most expensive top-of-the-line smartphone right before a cheaper, faster model drops, leaving you stuck trying to pay off an outdated asset.
13:22
Unidentified speaker (speaker_2): And to make that analogy even more accurate, imagine you bought that smartphone on a massive corporate credit line.
13:28
Unidentified speaker (speaker_0): Ouch.
13:29
Unidentified speaker (speaker_2): Right. If you buy a server cluster for $50 million,
13:32
Unidentified speaker (UNKNOWN): you
13:32
Unidentified speaker (speaker_2): likely finance that purchase through debt. Your entire business model relies on the assumption you can rent that equipment out. at a certain hourly rate for three to four years to pay back the bank and make a profit.
13:44
Unidentified speaker (speaker_0): So if a company like Cerebris introduces a cheaper, faster alternative into the market next year, you suddenly have to lower your rental prices to stay competitive with newer data centers.
13:55
Unidentified speaker (speaker_2): Yes. But your debt on the original equipment remains exactly the same. Aaron
13:58
Unidentified speaker (UNKNOWN): Powell
13:59
Unidentified speaker (speaker_0): But cheaper hardware overall can't be purely a negative for companies. whose entire business is buying and renting out hardware, right? Eventually they get to buy the cheaper hardware too.
14:09
Unidentified speaker (speaker_2): It is not entirely negative, which is why there is nuance to how this plays out over a long timeline. Cheaper computing could eventually improve profitability for these operators when they refresh their hardware.
14:19
Unidentified speaker (speaker_0): Right.
14:19
Unidentified speaker (UNKNOWN): If
14:19
Unidentified speaker (speaker_2): their input costs go down in the future, they can theoretically deploy more servers for less money and attract a wider customer base. Aaron
14:26
Unidentified speaker (UNKNOWN): Powell
14:26
Unidentified speaker (speaker_0): Okay, but what about right now?
14:28
Unidentified speaker (speaker_2): Well, outcomes strictly depend on financing, utilization rates, and customer demand right now.
14:33
Unidentified speaker (UNKNOWN): If
14:33
Unidentified speaker (speaker_2): an operator took on too much debt to buy the current generation of hardware, the transition to cheaper alternative architectures could be highly painful in the short term.
14:41
Unidentified speaker (speaker_0): Because their margins will compress while their debt obligations do not.
14:44
Unidentified speaker (speaker_1): Exactly.
14:45
Unidentified speaker (speaker_0): Okay. To briefly summarize our conversation for you listening, Cerebrus, raising its 2026. forecast and securing Microsoft as a customer highlights a clear set of winners in alternative chips and the manufacturing supply chain. At the same time, we see losers spanning traditional processors and infrastructure operators who are now facing strict depreciation and
15:05
Unidentified speaker (speaker_0): financing risks. Before we wrap up, what is one final thought you can leave the listener with?
15:09
Unidentified speaker (speaker_2): If alternative AI hardware continues to rapidly improve and drop in cost, what happens to the software ecosystem?
15:16
Unidentified speaker (speaker_0): Oh, that is a compelling question.
15:18
Unidentified speaker (speaker_2): Right. Will the barrier to entry for building AI models drop so low that computing hardware stops being the main bottleneck?
15:25
Unidentified speaker (speaker_0): Leaving the competition entirely to data acquisition and energy access. That is definitely something to mull over. This is not financial advice or buy or sell recommendations. This is purely the opinion of the author.
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.

More from Breaking News To Trading Moves

Breaking News To Trading Moves

Nvidia's Leadership Transition and AI Market Volatility

Episode discussion: Nvidia Leadership Transition and AI Market Volatility

Speakers’ claims: Speakers analyze Nvidia's market cap decline from $5.8 trillion to $3.9 trillion following CEO Jensen Huang's resignation and the appointment of interim CEO Colette Kress. They attribute the drop to multiple compression driven by management uncertainty rather than fundamental balance sheet deterioration. Options markets price in a 7% volatility move without directional bias, reflecting investor caution ahead of earnings. The episode identifies AMD, Broadcom, Microsoft, and Amazon as beneficiaries of customer diversification and hardware agnosticism. Conversely, hardware suppliers and leveraged infrastructure firms face risks from rising Treasury yields and potential demand shortfalls.

NvidiaAI MarketOptions Trading
17 min2 assets10 claims to verify
Breaking News To Trading Moves

Anthropic's $1.4 Trillion IPO

Episode discussion: Anthropic IPO mechanics drive supply-demand dynamics rather than fundamentals

Speakers’ claims: Speakers analyze Anthropic's planned $1.4 trillion IPO, noting that price action will be driven by mechanical index inclusion demand rather than business fundamentals. They highlight a scarcity dynamic where only 6% of shares are offered against an estimated $19 billion in passive buying pressure. The discussion details how hedge funds may front-run this demand while immediate selling pressure from unlocked private placement shares creates volatility. Strategic partners like Alphabet and Nvidia are validated by the deal, while competitors and other AI infrastructure firms face capital rotation risks.

IPO MechanicsIndex FundsAI Infrastructure
21 min2 assets10 claims to verify
All insights