20VC: Cerebras CEO on Why Raise $1BN and Delay the IPO | NVIDIA Showing Signs They Are Worried About Growth | Concentration of Value in Mag7: Will the AI Train Come to a Halt | Can the US Supply the Energy for AI with Andrew Feldman

Summary

Podcast Summary

Episode Title

20VC: Cerebras CEO on Why Raise $1BN and Delay the IPO | NVIDIA Showing Signs They Are Worried About Growth | Concentration of Value in Mag7: Will the AI Train Come to a Halt | Can the US Supply the Energy for AI — with Andrew Feldman

Podcast: The Twenty Minute VC (20VC)
Host: Harry Stebbings
Guest: Andrew Feldman (Co-founder & CEO, Cerebras)
Date: October 6, 2025

---

Episode Overview

In this episode, Harry Stebbings hosts Andrew Feldman, CEO and co-founder of Cerebras, to discuss the company’s historic $1.1 billion Series G fundraising round, the dynamics and future of the AI hardware and infrastructure sector, NVIDIA’s dominant role, the insatiable and unpredictable demand for AI compute, data center and energy challenges, and the geopolitical and societal impacts of the rapid AI revolution.

Feldman delivers frank, in-depth insights on market realities, AI chip innovation, talent wars, concentration of value among the “Mag7” tech giants, and the necessity for society to see real returns from vast investments in AI. The conversation is fast-paced, intellectually rich, and filled with memorable moments and actionable takeaways.

---

Key Discussion Points and Insights

1. The $1.1BN Raise: Timing, Investors, and Strategy

[04:04–06:22]

• Historic Nature: Cerebras’ $1.1 billion Series G at $8.1B valuation is the largest ever in their space, done at the highest valuation, and includes premier investors like Fidelity, Tiger Global, and Valor.
• Why Now?
  “We now have sort of the dry powder to really push…to build out our manufacturing to the scale and scope we want, to add new data centers…We have more big ideas."
  – Andrew Feldman [04:12]
• Signal Value of Investors: Fidelity’s participation is highly strategic — "they are the Oxford or Cambridge of investing…when they choose to lead a round, it brings Wall Street a great deal of confidence." [04:12]

2. The Decision to Stay Private a Little Longer

[05:53–06:22]

• Pre-IPO Round Logic:
  "It's very common in late stage to do a pre IPO round if you can get it done very quickly, if it doesn't distract you and keep moving."
  – Andrew Feldman [06:00]
• Retaining flexibility, seizing near-term opportunities.

3. The State of the AI Market: Hype, Exponential Demand, and Uncertainty

[06:22–09:34, 09:42–10:10]

• Breathtaking Uncertainty:
  “Things are moving at a rate that 6, 8, 12 months out. Everybody's unsure. It's so fast, it's so big."
  – Andrew Feldman [00:00, 06:40]
• Marketing vs. Reality: Feldman warns of inflated announcements (“up to $100B over 5 years”) and the need for realism about timeframes and actual value.
• Options, Not Commitments: Customer deals signal “options on the future” as nobody can accurately predict AI demand; companies are hedging rather than guaranteeing usage.
• Demand Underestimation:
  “100%. I've been wrong. If you would have said a year ago…that OpenAI would get the valuations they’re getting…it wouldn’t have been conceivable.”
  – Andrew Feldman [09:42]

4. Is the AI Boom Sustainable?

[10:20–11:15]

• Most AI experiments fail, but betting on “permanent Goliath dominance” is a zero-alpha game. Feldman’s bet: the economic pie (via AI) will get larger and reorganize society.
• NVIDIA’s Growth:
  “Of course, if Nvidia keeps growing at the rate they're currently growing, 11 years from now, everybody on earth works for them.”
  – Andrew Feldman [10:20]

5. NVIDIA’s Dominance and The Big Company Playbook

[11:35–13:07]

• Sign of Concern:
  “You use your balance sheet more and your technology less…you start buying business as opposed to winning business.”
  – Andrew Feldman [12:11]
• Tactics like predatory pre-announcements (e.g., B300s ahead of B200s) seen as evidence Nvidia is preparing for a more competitive future.

6. Decoding the OpenAI–NVIDIA $100B Deal

[13:07–14:53]

• Feldman asserts the deal structure was “designed for nobody to understand it”—deliberately vague and not truly analyzable.
• Analogy: “My mother goes shopping…I ask her, it's a lovely dress, Jules, how much is it? Well, it doesn't matter. It doesn’t matter."
  – Harry Stebbings [14:10]

7. Chip Depreciation, Performance, and the Physics of Progress

[14:53–17:57]

• Depreciation Reality:
  “People are clearly still getting value from H100s…and from A1000s…The question is: how much faster are future generations than the current generation? That's the actual question on depreciation.”
  – Andrew Feldman [15:17]
• Performance Increments: Marketing overstates improvements; real-world is typically “2-2.5x per meaningful generation move” [16:52].

8. SRAM, Memory Architecture, and Wafer Scale Innovation

[17:57–21:06]

• SRAM’s Limitations & Opportunity: SRAM is fast but small capacity – “your friend is exactly right…it's the reason we went to waferscale." [18:32]
• Cerebras overcame the 75-year “impossible” challenge of building dinner-plate-size chips – "what we were missing is that for 75 years nobody could do it…After we did it, Elon tried dojo and they failed…It was obvious, it was hard." [20:34]

9. Training vs. Inference: Market Split

[21:06–22:46]

• Cerebras Claims Leadership in Both:
  “No, we're faster on both. But the software challenges in training are real…in inference, nobody cares about Cuda, nobody even cares about Pytorch…put up a side by side to show you’re faster than 1000 B200s.”
  – Andrew Feldman [21:20]
• Higher friction to disrupt training workflows; inference is easier to switch and is a larger market.

10. The Exponential Growth of Inference

[22:46–23:54]

• “It’s really hard for the mind to wrap itself around geometric growth or exponential growth. More people are using AI…they use it more frequently…and what they want to do…is bigger.”
  – Andrew Feldman [22:52]
• All three factors are multiplying, driving staggering demand.

11. Productivity, AI Usage, and the Robert Solow Paradox

[24:08–26:13]

• Historical analogy to computers and electricity: real productivity comes only when society reorganizes around the new tech, not just replaces old processes.
• “If we use OpenAI the way we use Google, you’ll see a very modest jump in productivity. If you use them in a fundamentally different way…you’ll see a huge jump.”
  – Andrew Feldman [24:08]

12. Energy Demands: Is the US Ready?

[26:13–27:38]

• “People often say we don’t have enough power in the US and this is strictly wrong. We have plenty of power. It's in the wrong places."
  – Andrew Feldman [26:27]

• The challenge is a mismatch: power’s where people and fiber aren’t.

• Responsibility:
  “To the extent we consume this extraordinary amount of power, we have an obligation to deliver amazing things…If we use it and don't do that, then it's not a gain for society."
  – Andrew Feldman [27:38]

13. The Messiness and Market-Driven Exploration of AI Use

[27:51–29:06]

• Markets will fund wasteful, silly projects (“Ghibli images”), but sometimes what looks wasteful later seeds true advancement.

14. Geopolitics: US Policy, China, Immigration, and Energy Infrastructure

[29:06–31:03, 44:36–48:45]

• US Needs to do More:
  “We have real work to do…Our decentralized form of government has left us with a patchwork of power infrastructure…We have starved our universities of compute…”
  – Andrew Feldman [46:18, 48:17]
• Immigration:
  “We’re not making enough AI practitioners…Historically sucked the best and brightest on J1s and H1s…we need to do a better job.” [33:07, 47:23]

15. The “Mag7,” Market Concentration, S&P 500, and Risk

[31:03–32:44]

• “The risk is the mismatch in the mental model people have…they thought they were diversified and in fact they’re heavily dependent on a very narrow sector.”
  – Andrew Feldman [31:03]
• On NVIDIA:
  “They've proven themselves to be extraordinary company in the first quarter of the century. I don’t know if $4T is right, but maybe it’s too low.” [32:25]

16. Bottlenecks in AI Scale

[33:07–36:29]

• Expertise: Not enough AI/data talent produced by US universities; immigration policy makes it worse.

• TSMC & Manufacturing:
  “TSMC can’t build fabs fast enough…$30B, $50B factories…their ability to build them quickly enough is limited.”
  – Andrew Feldman [35:44]

• Data Center Capacity:
  “Where are those gigawatt facilities that everyone’s talking about? Well, they’re not up yet.” [36:29]

17. Data Centers and Investment Frenzy

[36:45–38:18]

• Data centers are a Wall Street favorite because they “look like a bond” (rent, stable tenants).
• “Building data centers is not for everyone…the ways to lose money in property are large and many…when you go unbelievably quickly it’s harder and harder to be disciplined.”
  – Andrew Feldman [37:43, 38:18]

18. Vertical vs. Horizontal Integration: Do AI Labs Need to Build Their Own Chips?

[38:18–41:50]

• Most successful AI labs (OpenAI, Anthropic) are not vertical – they use public cloud. Whether that remains true is an open question.
• On Chip-Building by Software Companies:
  “There is a long history of software companies failing to build chips…the answer is this is really hard…mentality is very different—it’s not ‘move fast and break things’.”
  – Andrew Feldman [39:26–41:50]

19. What Will the AI Chip Market Look Like in 10 Years?

[42:03–42:39]

• “Absolutely not one takes 90%…even at Intel’s strength they had dominance in x86 and 0 market share in cell phones…It will not all accrue to one or two companies.”
  – Andrew Feldman [42:10]

20. Margins, Pricing Power, and the Coming Shakeouts

[42:43–43:50]

• Cerebras raised more due to positive margins; NVIDIA’s margins are “extraordinary”—as high as 78–85% for certain products.
• “When the giant stumbles, the number of people who come out of the woodwork to kick them when they're down is extraordinary…it was years of pent up frustration…”
  – Andrew Feldman [43:12]

21. Sovereignty as an AI Business Model

[43:50–44:36]

• Feldman sees regional sovereignty (e.g., Mistral in Europe) as a viable way to gain traction when combined with real product advantage.

22. The US–China AI Race

[44:36–46:18]

• “I think it benefits neither…we would be much stronger if we can find ways to peacefully engage…their government has an extraordinarily aggressive policy in AI…we have real work to do.”
  – Andrew Feldman [44:57]

23. The War for Talent

[34:14–35:41]

• “No company ever went bankrupt by paying extraordinary people too much. If you want to go bankrupt, pay mediocre people too much.”
  – Andrew Feldman [35:13]
• The compensation for elite technical talent is rational due to their “unreplicable” impact.

---

Notable Quotes & Memorable Moments

Memorable Quotes

• On Planning for Uncertainty:
  "You plan more frequently, you have a shorter view, you take options on the future and, if the future moves against you, you lose the premium on the option."
  – Andrew Feldman [08:39]

• On Market Hype:
  "The great sort of CYA word in marketing history is ‘up to 100 billion over five years'.…You could pick a lot of big numbers and it won't be bigger than."
  – Andrew Feldman [06:40]

• On Software Companies Building Chips:
  "Modern software does not fit well in a chip making framework…Weekly sprints don't work well on two year long projects."
  – Andrew Feldman [39:26]

• On AI’s Societal Obligation:
  "If we are going to consume this amount of power, the burden is on us to deliver value for it."
  – Andrew Feldman [27:38]

• On Talent:
  "There are engineers who have skills that no number of other engineers working together can achieve…It's why the best and the brightest are getting such extraordinary compensation."
  – Andrew Feldman [34:25]

• On Leadership Challenges:
  "Every day I go to battle with Goliath…Every dollar we sell is a dollar that if we didn’t work at it, …would default to Nvidia."
  – Andrew Feldman [59:49]

Humorous, Relatable Moments

• On the Demand for AI Compute:
  "We have customers coming to us and saying we would like between 5 and 40 million queries per second. Well, how do you not know by a factor of 35 million queries per second?"
  – Andrew Feldman [06:40]

• On Market Announcements:
  “My mother goes shopping and I ask her, it’s a lovely dress, Jules, how much is it? Well, it doesn’t matter. It doesn’t matter.”
  – Harry Stebbings [14:10]

• On The First Cerebras Wafer-Scale Chip Succeeding:
  "...founders stood there together and stared at the box running, which is about as interesting as watching paint dry. And we stood there and we couldn't believe it."
  – Andrew Feldman [53:00]

---

Timestamps for Important Segments

• 04:04 – Why raise $1BN? Why now?
• 06:22 – State of the AI market, demand uncertainty
• 08:39 – Planning amid exponential uncertainty
• 10:20 – Sustainability of the AI boom
• 11:35 – NVIDIA’s evolving competitive strategy
• 13:07 – Analyzing the OpenAI–NVIDIA partnership
• 15:17 – Chip depreciation and generational improvements
• 18:32 – SRAM, architecture, and reasons for Cerebras' wafer-scale innovation
• 21:20 – Training vs. inference, Cerebras' comparative performance
• 22:46 – The exponential growth of AI inference demand
• 24:08 – Productivity, computing, and historical analogy
• 26:27 – US power supply and energy for AI
• 31:03 – Concentration risk in the S&P/“Mag7”
• 33:07 – Key supply and talent bottlenecks
• 39:26 – Why software companies struggle (and often fail) in chip design
• 42:10 – The long-term landscape for silicon/AI chips
• 43:08 – NVIDIA's margins and impact on industry structure
• 44:57 – Geopolitics, China–US competition, and relevant policy
• 46:18 – US weaknesses: immigration and infrastructure
• 53:00 – “Moment of victory” for wafer-scale chip
• 54:52 – Where should investors look next? Data pipelines!
• 56:42 – Will AI create labor shortages? Feldman's skeptical take

---

Conclusion & Takeaways

Harry and Andrew pack this episode with gems for founders, investors, policymakers, and technologists:

• Understand the difference between hype and operational reality.
• Appreciate the nonlinear, option-laden nature of AI demand and the importance of agility in planning.
• NVIDIA’s dominance is real, but the playbook of giants contains the seeds for future disruption.
• Data center build-out, manufacturing, and talent remain the real bottlenecks for scaling AI.
• Societal returns must justify AI’s massive energy and economic footprint.
• The future belongs to those with patience and a fundamental understanding of both tech and markets—not just those following momentum.
• AI’s adoption curve will mimic that of other fundamental technologies: real transformation comes only when we reorganize around it, not when we simply replace old tools.

---

Further Listening & Resources

For more on this and similar topics:

• Visit 20VC.com
• Search for previous founder episodes with guests like Daniel Ek (Spotify), Reid Hoffman (LinkedIn), and Frank Slootman (Snowflake).

Transcript

[Andrew Feldman]

Things are moving at a rate that 6, 8, 12 months out. Everybody's unsure. It's so fast, it's so big. There is unbelievable demand and nobody knows where it will go in the future. The question of depreciation is how much faster are future generations than the current generation? That's the actual question on depreciation. People often say we don't have enough power in the US and this is strictly wrong. We have plenty of power. It's in the wrong places. Risk comes in financial markets where people fundamentally underestimate risk. No company ever went bankrupt by paying extraordinary people too much.

[Harry Stebbings]

This is 20 VC with me, Harry Stemmings, and I'm so excited for the show today following our blockbuster episode with Jonathan Ross at Grok last week. I'm so excited to welcome another leader in the space today in the form of Andrew Feldman, co founder and CEO of Cerebras, building the world's fastest AI inference and training. Now, Cerebras recently closed a $1.1 billion Series G round at an $8.1 billion valuation. With names like Fidelity, Valor and others included in the round, they've leapfrogged GPU limits, operated trillions of tokens per month, and are filing to go public very soon. This was an incredible discussion. I'm so grateful to Andrew for his friendship and I hope you enjoy the show. But before we dive into the show today, I love seeing the team come together to make this show happen. What I don't love is trying to keep track of all the information, the data and the projects that we're working on across dozens of platforms, products and tools. That's why we use Coda, the All In One collaborative workspace that's helped 50,000 teams all over the world get on the same page. Offering the flexibility of docs with the structure of spreadsheets, Coda facilitates deeper teamwork and quicker creativity and their turnkey AI solution. The intelligence of Coda Brain is a game changer. Powered by Grammarly, Coda is entering a new phase of innovation and expansion, aiming to redefine productivity for the AI era. Whether you're a startup looking to organize the chaos while staying nimble, or an enterprise organization looking for better alignment, Coda matches your working style. Its seamless workspace connects to hundreds of your favorite tools, including Salesforce, Jira, Asana and Figma, helping your teams transform their rituals and do more faster. Head over to Coda iO20VC right now and get six months off the team. Plan for startups for free. That's Codacoda iO20VC and get six months off the team plan for free. Coda iO20 and talking about trust today customers expect it faster than ever. And that's why over 10,000 global companies trust Vanta. Vanta automates up to 90% of the work for in demand compliance standards like SoC2, ISO 27001 and more. Using smart AI to centralize workflows, manage risk and get you audit ready in weeks, not months so you can stop chasing paperwork and start closing deals. And a new IDC report found that Vanta customers achieved $530,000 per year in benefits. That's insane. And the platform pays for itself in three months. I had no idea about these. Whether you're growing fast or just getting started, Vanta connects you with trusted auditors and experts support to help you build trust with customers. Get a thousand dollars off your first [e-mail] 20VC. That's vanta.com 20VC.

[Andrew Feldman]

You have now arrived at your destination.

[Harry Stebbings]

Andrew, dude, it is so lovely to have you back on. I so enjoyed our first show. You put up with my naive questions enough to agree to do a round two. Man, I must be charming.

[Andrew Feldman]

Harry. I'm okay with any questions, naive or otherwise. Happy to do it anytime. I read your LinkedIn posts, your Twitter posts. I'm rooting for your mom.

[Harry Stebbings]

Dude, you are too kind.

Listen, I want to start with the.

Billion dollar raise that you just announced yesterday. Can you just talk to me? Why?

It's important.

Why now? And what it means for the look.

[Andrew Feldman]

It was the largest raise ever done in our category. It was done at the highest valuation and with the premier investors at late stage investing. You're looking for the likes of Fidelity. They are the, what would the English call it, the sort of Oxford or Cambridge of investing. Right? I mean they are the premier public market investors and when they choose to lead a round, it brings Wall Street a great deal of confidence. We were really happy to partner with them and with the treaties to lead the round. And then we were able to get enormous participation from Tiger Global, from valor from 1789. So that's point one. I think point two is that we now have sort of the dry powder to really push take the opportunities in front of us to build out our manufacturing to the scale and scope. We want to add new data centers. We added five this year in the US to add more data centers and we have more big ideas. I think incremental improvements make believe gains achieved by dropping from, you know, 8 bit to 4 bit. Those aren't going to get us to the promised land in AI. We've got real work to do as a community. I think this funding puts us in the catbird seat for that.

[Harry Stebbings]

On the fidelity side, it's actually interesting. I have Brian Halligan, the CEO of HubSpot on the show recently and he actually taught me the importance of specifically getting fidelity in both pre IPO and when you IPO just because of the signal that it sends. And I didn't actually realize as a venture guy the weight that's placed as.

[Andrew Feldman]

An early stage venture, your guy, you're like, who are the public guys?

[Harry Stebbings]

Okay, Fidelity, like whatever. T row, sure. They're all the same, right?

And I was like, no, no, they're not like fidelity are the monster in the room.

[Andrew Feldman]

They're.

[Harry Stebbings]

The importance of getting them is. Is very high.

Can I ask you, dude, like, why not go public? Because it was rumored that you guys.

Were going to go public. Why do this pre public round?

[Andrew Feldman]

We still have every intention of going public. I think it's very common in late stage to do a pre IPO round if you can get it done very quickly, if it doesn't distract you and keep moving. I think there were so many opportunities in front of us that gathering the capital so that we could continue to prosecute these opportunities was a no brainer.

[Harry Stebbings]

You said about the real work to be done. I think it's quite difficult for everyone who's not really in the market to understand what the hell's going on. Given all the news that we see. Can you help us just with the lay of the land in the last three months of where are we at now? What's changed?

[Andrew Feldman]

The first thing, Harry, is we are in a stage of the market where the claims are enormous, where tens of billions of dollars are being done here and there and nobody's reading the fine print that it's over five years and it's up to. This is the great sort of CYA word in marketing history is it will be up to 100 billion over five years, right? Well, up to means it could be 30, it could be 12, could be 40, right? He could pick a lot of big numbers and it won't be bigger than. And so as you read these deals, I think you have to really think about the timeframe over which they're being done. You have to think about whether anybody is actually counting. Lots of people are saying they're gonna bring hundreds of billions of dollars of jobs to the US and this and that in eight months. Has anybody got a little spreadsheet like nine jobs plus one factory. I mean, who holds anybody to account? And the answer is nobody. I think that's number one. I think number two, what this signals more than anything is that there is unbelievable demand and nobody knows where it will go in the future. That it's so big and happening so quickly that they don't know. We have customers coming to us and saying we would like between 5 and 40 million queries per second. Well, how do you not know by a factor of 35 million queries per second where your demand's going to be? How are you unsure by an order of magnitude of what your queries are going to be? The answer is things are moving at a rate that 6, 8, 12 months out. Everybody's unsure. It's so fast, it's so big. And so you should think about sort of these announcements as sort of options on the future. Right. That's really the way to think about it is in an unknown environment, how can I take an option on the future? I don't know if I'll use it all, but I'll pay something for the future rights to have some capacity. So that's a way to think about it.

[Harry Stebbings]

Given that it's so fast, it's so big, how do you think about planning for that uncertain future?

[Andrew Feldman]

It's brutal. There's a very interesting question about in extraordinarily rapidly moving environments what the right planning cadence is. What you really need is good planning changing rules rather than good planning. Right. We have to make big bets. We're making five and seven year investments in data center capacity. We are making hundreds of millions now in terms of billions of dollars of bets in supply chain. And those are not three month bets. I think what you need to do is use different rules that have historically been used. You plan more frequently, you have a shorter view, you take options on the future and if the future moves against you, you lose the premium on the option. You pay a little price to secure some capacity and if you don't use it, you just, you go, all right, that was sort of a way to manage uncertainty about the future.

[Harry Stebbings]

What do you think the chances are that you are still underestimating even your wildest of demand expectations?

[Andrew Feldman]

100%. I've been wrong. If you would have said a year ago, two years ago, pick a time at which it would have been conceivable that OpenAI would get the valuations are getting. It wouldn't have been conceivable three months ago, six months ago, nine months ago. When was it like the day before. Right. I think that's true with the demand we're seeing. It's true with the valuations on companies we're seeing. It's true with the rate of ideas entering the community.

[Harry Stebbings]

How much of this do you think is sustainable? Everyone lobbies the oh, it's not sustainable. A lot's experimental, it's not enduring. How much do you think is sustainable?

[Andrew Feldman]

I would say they're always grumpy people who like, it'll never work and you'll never beat Goliath. And truth is, most things don't work and most of the time Goliath wins. But there's no alpha in that. There's no money made for you or me. Betting on the biggest of the big dogs to continue not to lose. How uninteresting is that? Of course, if Nvidia keeps growing at the rate they're currently growing, 11 years from now, everybody on earth works for them. Do the math. However, is it possible that our economy looks very different in five years? Is it possible that the things we value are very different? That we have reorganized around AI, we've seen a major bump in labor productivity, we have benefited dramatically and the economic pie is much larger. I think that's not only likely, it's almost certain.

[Harry Stebbings]

You mentioned that if Nvidia continues to grow the way that they do, everyone will work for them.

[Andrew Feldman]

You keep doubling at that rate, you multiply. You can't keep doing that.

[Harry Stebbings]

To what extent is it just completely unshakable at this point for them where the scale and the size of money is like Jonathan Ross from Grok said on the show, they will unwaveringly get $10 trillion within a five year timeline.

[Andrew Feldman]

I hope he's long on them then. I don't pick public market stocks. I don't like picking public market stocks. In the public market you can lose money on good companies, you can make money on shitty companies. And that for me doesn't sit well as an entrepreneur, as a David in the battle with Goliath, I want to make money when we build a great company, period. But can they continue to grow? I think we are seeing some things that big companies do as they begin to worry about growth. I think use your balance sheet more and your technology less. This is something that historically large companies have done as they feared for their technical prowess.

[Harry Stebbings]

And when you say that, you kind of referring to investments in your OpenAI of $100 billion and 11 labs and.

[Andrew Feldman]

Everyone in between, you start buying business as opposed to winning business. We saw that with Cisco, who emerged as a dominant position 99, 2000, 2001. That has been one of the strategies. Another strategy you see is this predatory preannounce where you announce B3 hundreds before anybody can get B2 hundreds. You start talking about Rubin before B2 hundreds are technically finished. You don't talk about the field failure rate of your products, which are massive. Rather, you sort of keep talking about the future in an effort to convince people to wait to make a good decision rather than go with technology that's better and present. I think these are the strategies of very large companies using their strengths. And I think that's what you're beginning to see unfold with Nvidia.

[Harry Stebbings]

Speaking of the 100 billion into OpenAI, how did you analyze that? For me reading that, I didn't really know how to analyze it. It's so unprecedented.

[Andrew Feldman]

It was designed for nobody to understand it. If one wants to make something very clear in an investment, we've invested this amount at this valuation, the deal's done. Now, if you want to make something more difficult, it's up to this amount over an unbound specified amount of time at no valuation given or evaluation specified. But it can change. It wasn't designed for you or other analysts to anchor on different things. And that's a very reasonable thing for both of them. But it's just, it's not an analyzable thing. Beyond the fact that Nvidia has chosen to try and lock up a portion of OpenAI's demand by investing in them, that's about as much as you can.

[Harry Stebbings]

Say totally get you. And I'm glad that it's meant to be confusing because I was confused looking at it going, what price was this?

How much are they buying?

[Andrew Feldman]

Like, I don't know if it was meant to be confusing. I think it was meant to be.

[Harry Stebbings]

You know, my mother goes shopping and I ask her, it's a lovely dress, Jules. How much is it? Well, it doesn't matter. It doesn't matter.

[Andrew Feldman]

You call your mom. Wait a sec. Let's go back to the important thing. You call your mom by her first name?

[Harry Stebbings]

Oh, yeah, Jules.

[Andrew Feldman]

Okay. You don't call her Mom. I've never called my mom Shirley. I mean, I. I could never call her. I mean, that was just mom or something else. But not.

[Harry Stebbings]

No, no. But when I, when I get the, like, price on application, I'm like, oh, Andrew.

[Andrew Feldman]

Oh, yeah.

[Harry Stebbings]

Oh, that's what this felt like.

That's what I was like. Really?

[Andrew Feldman]

No, nothing that. The truth is, is there may be and they're likely a huge number of moving parts that make it impossible to clearly describe without giving out more than they wanted to give out.

[Harry Stebbings]

You mentioned pre announcements that like B3 hundreds, B2 hundreds. Timings of such. Are we thinking about chip depreciation in the right way again? I just did a show with Jonathan, he's like, hey, we actually think about them on like an 18 month time cycle to maybe two years. And I was like, wow, it's quite quick. Are we thinking about it the right way? And how should we be thinking about the amortization of chips?

[Andrew Feldman]

We are in unprecedented waters. People are clearly still getting value from H1 hundreds and that's more than two years, right? So if you say it's a two year depreciation, you're empirically wrong. And I think people are still getting value from a 1000s, though not on the cutting edge. And so that's closer to three or four years and could be as long as five or six. The question of depreciation is how much faster are future generations than the current generation? That's the actual question on depreciation. Because with depreciation you're saying at some point it's no longer worth using a part that's fully paid off because there's a new part that's so much faster, uses so much less power, that it's better for me to retire it. That's the actual, the underpinning to the depreciation question. If I have a data center and it's 50 megawatts and I have this much capacity in it, at some point, even though my chips in it have been depreciated and I'm running them at zero cost, right? Power plus zero depreciated cost, it makes sense to move them out because the new chips are so much faster, so much better, use so much less power, I get so much more dollars per. And so that's the question. If we don't as an industry continue to build extraordinarily sort of better parts generation after generation, then people don't move from one generation to the next. They last longer. You depreciate them longer.

[Harry Stebbings]

Where are we? I feel very naive for asking this. Where are we in the performance improvement pathway for chips? Are we like in there? We've got 90% and we're at incremental gains or are we at the, we are still at the super early stage and we have 90% of the gains to be made.

[Andrew Feldman]

The question is in that case whether you read people's marketing material or the actual performance results. Certainly people's marketing material would lead you to believe that generation over generation, they're huge gains. A little bit of engineering digging probably leads you to the conclusion that you're getting 2.2.5x per meaningful generation move, not more. If you Compare apples to apples, 8 bit to 8 bit, 4 bit to 4 bit, if you compare actual performance, you might have more flops on the chip, but your memory bandwidth didn't improve more than 2x, so you can't get to them. These chips are a solution, and if you make one part fast, it's a system. You make one part fast and the other part doesn't move as far forward, it becomes the new bottleneck. It doesn't matter how many flops your chip has, if you can't get data onto and off of the chip, those are wasted. The question isn't how much, how much faster is the chip, it's how much faster is the solution. That includes memory, which has, for inference, is the fundamental limiter for the GPU architecture. And so it doesn't matter how much faster the chip goes, it matters how much faster the memory bandwidth is.

[Harry Stebbings]

On this, I was chatting to a founder in the space and he said that what everyone fails to understand is that although SRAM sounds great in terms of having memory, and SRAM is obviously, you'll describe it much better than me and hate me for this, but SRAM is obviously memory on chip versus off chip, seemingly great, but he said it's completely unable to handle scale. And so although it may be quicker for anyone who wants to do large scale, it is incapable at present of doing that. And that's a fundamental need and requirement of any of the large providers. Do you think that's fair and how do you think about that?

[Andrew Feldman]

Not only is it fair, it's the reason we went to waferscale. What your friend said is strictly true in that SRAM is blazing fast and low capacity. HBM is a flavor of dram. It has high capacity and it's very slow. Now, Nvidia and all GPUs, including AMDs, chose a big capacity memory that is slow because it's perfect for graphics. You don't have to go to memory very often. You can hold a lot, you don't go very often. SRAM is blazing fast, but it can't hold very much. So the problem on traditional chips is if you put memory on the chip, you are using space that could be otherwise used for compute. You have a fixed amount of real estate. And so if you Put half memory, half your real estate's available for compute. Our idea was that if we sort of, if we built a chip that was the size of a dinner plate, we could stuff it to the gills with fast sram. Overcoming the limitation of sram, which is it doesn't store very much by putting a huge amount down by using more silicon area. Now if you're an SRAM solution today in a normal sized chip and you're trying to do a trillion parameter model, use 4 or 5000 chips, what a mess. You know how many cables that is? Do you know the impact to the AI? It's a horrible mess and it limits you from doing things you want to do with the AI, like speculative decode. It has all sorts of painful challenges. The other hand, use one of these or two or four and it's simple, it's easy. And this is what your friends said exactly right. And the reason we went to build a bigger chip so we could fill it with this fast sram, so we could get over the traditional limitations of SRAM that couldn't store very much by using a lot of space, by using a huge amount of silicon area. So your friend is exactly right. You might listen to him again in the future.

[Harry Stebbings]

Question to you, no offense, that seems a little bit obvious, like, okay, increase the real estate, shove more SRAM on.

[Andrew Feldman]

It does, doesn't it? Yeah.

[Harry Stebbings]

Is it as obvious as it seems? Am I missing something here?

[Andrew Feldman]

Well, what we were missing is that for 75 years nobody could do it. Building a bigger chip had proven impossible before we did it. Nobody in the history of the computer industry had been able to build a chip bigger than about 840 square millimeters in the 75 year history the compute industry. And many people had tried and failed. After we did it, Elon tried dojo and they failed. It's really, really hard. Our strategy had never been done before, never been successfully yielded. And so while it was obvious, it was hard.

[Harry Stebbings]

So when we think about where the market is today in terms of training and inference, do we agree then that actually Nvidia's chips are much better for training than yours are, but yours are much better for inference than theirs are and the market splits in that respect?

[Andrew Feldman]

No, we're faster on both. But the software challenges in training are real.

[Harry Stebbings]

What does that mean?

[Andrew Feldman]

It means that when a new model is built and everybody reads about it in a publication, it was done on a GPU for everybody to train it, they take the recipe that was originally done for the GPU and they have to Move it to the recipe for their hardware. Whether that's a tpu, whether that's an AMD gpu, whether that's another dedicated chip like ours, you have to move it. And that's a harder software lift. In inference, the truth is nobody cares about Cuda, nobody even cares about Pytorch. What they want is an API. It's literally 10 keystrokes to move from a GPU based solution on OpenAI OSS120B to our solution. It's 10 keystrokes. That's it. It's nothing. The answer is that while we are faster at training and we are faster at inference, it's easier to demonstrate inference. Right? You just put up a side by side to show that you're faster than a thousand B2 hundreds. You got to get 1,000 B2 hundreds. You need to train the model for four weeks or six weeks. You got to stand up a cluster of our machines. It's a bigger lift. I think the market right now is easier to move people off GPUs in inference. And the number of people doing inference is vastly higher than the number of people doing training.

[Harry Stebbings]

When you look at the inference market today, how has it developed in a way that you did not expect?

[Andrew Feldman]

I think it's really hard for the mind to wrap itself around geometric growth or exponential growth. I think there is nothing confusing about the rate of growth of inference. Rate of growth of inference is the number of people who use it times the frequency of use times the amount of compute needed per use. It is three different variables multiplied by each other. The problem is they're all growing fast and that produces some mind numbing effects. More people are using AI. Once they start using AI, they use it more frequently and what they want to do with it is bigger and more complicated. So it uses more computers. And so you have three variables. The size of the markets, the product of the three, all growing fast. We knew that going in. We see that and it still takes your breath away. That is really, really interesting.

[Harry Stebbings]

I think we've seen anything yet.

[Andrew Feldman]

I agree. I think the reason I am 100% sure that we are underestimating the market is because of that premise.

[Harry Stebbings]

I think Sam Altman said it very well in terms of how people use ChatGPT, but he said essentially the majority of people use it like Google, but a Google replacement. And actually the younger people use it as an operating system for the future, which is the right way to do it in his mind.

[Andrew Feldman]

Absolutely right. I think in 1988, Robert Sallow, who won a Nobel Prize in economics. He asked this question. He said, we see computers on every desktop and everywhere we look, except in the productivity statistics. And you say, well, that's a really interesting thing to say. And there was another economic historian who jumped into the fray. And he did this huge study. His name was Paul David, and he wrote a very famous paper called the Computer and the Dynamo. And what he studied was the adoption of electricity in the manufacturing sector between about 1880 and 1955. And what he showed was at the beginning, electricity produced very little productivity gains. It was basically used as a backup for belt driven systems. And it wasn't until they reorganized the shop floor to take advantage of electricity that you got this huge jump in productivity. And if you roll that forward to the computer, what he was saying is, look, we used computers to do things we were already pretty good at. We replaced a typewriter, we replaced general ledger accounting with spreadsheets. We were good at those things. You didn't get a big jump. And what he predicted and then happened immediately thereafter, by the mid-90s, you had a huge jump in productivity. We had begun tying them together. We'd built the Internet, we had the first parts of a cloud, and all these things used compute differently in ways that had never been consumed before. And you got this massive jump in productivity. If you use OpenAI in various of their competitors, the way you use Google, you'll see a very modest jump in productivity if you use them in a fundamentally different way. That was Sam's point. You'll see a huge jump. And if we reorganize ourselves around AI, you're going to see massive productivity gains. If we use AI to replace things we're already doing, Google or something else, you're not going to see very big jumps at all. And so that transition takes time. And what he pointed to was a demographic. Younger users are using it in a different way than older users. Older users are replacing something they already had. Younger users are using it in a way that never existed before. An operating system for life.

[Harry Stebbings]

If we're going to see that transition that you mentioned there, the energy requirements are just insane. I mean, Sam said a trillion dollar spend. He needs the energy of Japan.

[Andrew Feldman]

Yeah.

[Harry Stebbings]

Is this feasible, Andrew?

[Andrew Feldman]

Yeah, it's feasible. Is it desirable or good for society is a different question. It's feasible. People often say we don't have enough power in the US and this is strictly wrong. We have plenty of power. It's in the wrong places. Right. It's not where we have people or where we have Fiber optic cable. We have a ton of power in West Texas in natural gas. We have a ton of power in in upstate New York, in hydro. We have a ton of power in lots of places. We don't have people there. The problem is one of a mismatch between where all the power is and where the people are or where the buildings are or where the telco fiber is that we need to get data to and from the data center. The second observation is one of a community and that's that to the extent we consume this extraordinary amount of power, we have an obligation to deliver amazing things. And that's on all of us. I think we have an obligation to deliver drugs that are more efficacious, to deliver better health care, to make aging less painful, make the looking after of aged parents or sick parents. You go through society's ills and woes. If we are going to consume this amount of power, the burden is on us to deliver value for it. If we use it and don't do that, then it's not a gain for society.

[Harry Stebbings]

Do you think that's controllable? You know, creating Ghibli or Ghibli images, I get it wrong, isn't particularly value inducing, but it's churns a huge amount of compute and energy. Can we control that?

[Andrew Feldman]

It's a very hard question and I'm one voice in this. The problem with markets is that they do a lot of things that aren't productive in order to get one that is very productive. Ghibli may or may not have been net societal gain, but maybe the technology that is used in Ghibli is used for X ray crystallography and later is fundamental to finding major scientific breakthroughs. I mean that's the messiness at any given point in time, right? You can point to in a thousand poppy strategy, which is what a market is, right? A market has a lot of bad ideas. To get a few good ones, that's your business. Your business is investing behind a lot of big ideas, most of which fail. And the market is that writ large. And so in that environment you can always point to oh, that was a dipshit investment, Harry. Why do you invest with them? I mean they blew up. You can always say that after the fact. Look at that. They're using a ton of energy for that. That's not useful. Let's see. And so I think the answer is we need to be sure that at a societal level, where we use government dollars, where we use tax breaks, where we use permitting breaks. We are sure that we are giving these to disproportionately to projects that matter to society.

[Harry Stebbings]

Do you think Trump's done more to help or to hurt the US AI?

[Andrew Feldman]

I think it's confusing on net. It's probably more to help. The Biden administration was misguided and afraid the Trump administration. I think to his credit he surrounded himself with some smart people in the AI space. On net it's been positive.

[Harry Stebbings]

When you look at what is required in terms of energy, is nuclear unavoidable or the sole solution to be the providing force of energy for this next generation of AI?

[Andrew Feldman]

No, it's not unavoidable. It's a very reasonable decision for countries that don't have lots of alternatives. Canada has more falling water than anywhere else on earth. The opportunity for Canada to develop the cheapest power on earth is mind boggling. There is cheap power in lots of places. But in order for countries that wish to and don't have the natural resources of Finland that has geothermal, or Iceland that has geothermal, Canada that has falling water, nuclear is a very reasonable and cost effective strategy, especially over a several decade view.

[Harry Stebbings]

What worries you most today, Andrew?

[Andrew Feldman]

I do think of this idea that to consume the resources we're consuming, we have to be sure that we produce some extraordinary outcomes. I worry that the opportunity is so big that as a community we're running sort of helter skelter at it, that actually sometimes instead of running where you trip and fall and graze your knee and chip a tooth, if you stopped and thought and marched, you might get further over a 30 or 60 or 90 day period.

[Harry Stebbings]

Do you worry about the concentration of value in Mag7? They now make up more of the S and P than they pretty much ever have done in history. And that concentration in value is very real. If AI hits a speed bump in any way, the market could derail significantly and the multiplier effect of that is felt by everyone.

[Andrew Feldman]

Right. I think the risk there is not that they consume that much, that they are that much value. That's not the risk. I think they're that much value because we believe the future economy will reward that. I think the issue is that people then think the S and P is a safe investment or a safer investment investment than it might be. The risk is the mismatch in the mental model people have. Risk comes in financial markets where people fundamentally underestimate risk. When risk is priced properly, then what happens is your outcomes are not surprising. If people continue to think the S and P is Sort of an index of the global economy. And it's not, it's 30% or 50%, seven companies. Then they're exposed to sector risk that they weren't signing up for. They thought they were diversified and in fact they're heavily dependent on a very narrow sector. And that's a risk that seems to me to be a challenge in the new world order. And all the advice that sort of the pundits give a diversified portfolio when the world changes and that portfolio is not. You keep holding it and it's not diversified anymore because of consolidation, then there's real risk.

[Harry Stebbings]

Do you think the risk is priced properly? When you look at Nvidia at four and a half trillion, I think they've.

[Andrew Feldman]

Proven themselves to be the greatest company of the first quarter of the 21st century. They've proven themselves to be extraordinary company in the first quarter of the century. And I don't know if 4 trillion is right, but I think a very big number, maybe it's too low, is right because of what they've achieved.

[Harry Stebbings]

When we look at, we've said before about the insatiable demand that we cannot predict or anticipate, what are the bottlenecks today in your mind? Again, we had Jonathan at Grok on and he was like, actually, you know, I had someone come and demand five times the supply that I have in total. That was from one customer. Supply is mine. How do you think about the bottlenecks that we have to reach the insatiable demand that you mentioned?

[Andrew Feldman]

I think if you go back to planning, if you've got customers demanding 5x your capacity, I mean, probably didn't get your planning right. Right. You probably should have, should have planned better. I think there are bottlenecks at every level that are real and meaningful. I think the first one is expertise. We have fundamental limitations in AI expertise. We're not making enough AI practitioners. We're not making enough data scientists who understand data pipelines. Our universities aren't minting enough. Our challenges in the US with immigration don't help that. We have historically sucked the best and the brightest first on J1s to come to our schools and H1s to stay. If that is not our policy, we need to make them. If the government decides that that is not the way they want to build a workforce. Instead they want to build it out of people who live here. We need to do a better job of training those people. We need to do a better job of teaching them in K through 12. We need to do a better job of educating them in our universities in order to make the number of engineers we need to meet this demand. That's a bottleneck and it's why the best and the brightest are getting such extraordinary compensation.

[Harry Stebbings]

Is the war for talent completely out of control? You're seeing your zucks of the world spend hundreds of millions on one person. Do you think that's a blown up anomaly or do you see the war for talent being unprecedented?

[Andrew Feldman]

There are engineers who have skills that no number of other engineers working together can achieve. There are scientists who do, who have ideas and who have brains that are. It can't be replicated by lots of other talented people working together. Ought they to be paid more than world class soccer players? I have no idea. Maybe, maybe not.

[Harry Stebbings]

I mean inherently, yes, from an economic rationale standpoint, yes. The value generated from a Chief Scientist at OpenAI. If they add $50 billion of enterprise value to pay them, a billion dollars is worth it.

[Andrew Feldman]

That's what we have to think about. I mean, we've paid Charlie Sheen two and a half million dollars an episode for Two and a Half Men. I'm pretty sure that there are lots of people whose net productivity to society is above that.

[Harry Stebbings]

And he still spent it all.

[Andrew Feldman]

I just saw the show. Was it Netflix or Prime? What a sad story of somebody who was so self destructive and so talented. But should we be paying soccer players or basketball players? I have no idea. And I don't spend a minute worrying about whether we're paying extraordinary people too much. No company ever went bankrupt by paying extraordinary people too much. If you want to go bankrupt, pay mediocre people too much. That's how you mess up. Nobody's ever struggled by paying truly extraordinary people too much.

[Harry Stebbings]

What's the other bottleneck? You said expertise is one.

[Andrew Feldman]

TSMC can't build fabs fast enough. The truth is that these are both for TSMC and Samsung. These fabs are the most amazing manufacturing plants on the planet. You know, these are $30 billion, $50 billion factories. Their ability to build them quickly enough is very much limited. That in turn limits and keeps the supply below where it would like to be of chips. Not just our chips or Nvidia's chips, but everybody's chips. Below where it might otherwise be keeps the cost up. Right now there's a shortage of data center capacity. There's a huge amount of investment that has gone into that. There's a lot of words, but where are these gigawatt facilities that everyone's been talking about? Everybody's committing to them. Where are they well, they're not up yet.

[Harry Stebbings]

How long does that take?

[Andrew Feldman]

You know, somebody like for Elon, who's the fastest in the world and maybe the best at building plants and large construction projects, it takes six months, eight months, and for the rest of the world it takes a year and a half, maybe longer.

[Harry Stebbings]

Are we investing enough in data center builder? I mean, it is one of the most insanely hot categories now in terms of investment properties. I'm coming from like a pure Wall street mindset. Like every Wall street guy wants to be in data centers, it has a.

[Andrew Feldman]

Structure that they really understand, right? It looks like a bond to them. It looks like a piece of real estate. You get a tenant, they pay rent every month, you can loan against that. You get an investment grade tenant that's based on basically a bond. It has the advantage of falling into a category or a pattern that is really well understood in the debt market and in the capital markets. And that's an advantage. Coreweave and some of their sort of financial engineering and innovations there help the world see that like many things, lots of people will enter. The smart will make money, the less sophisticated will lose money. Building data centers is not for everyone.

[Harry Stebbings]

How will you lose money building data centers?

[Andrew Feldman]

I think if the best can build them for 8 million a megawatt and you're spending 12 or 14, that's how you lose money. You lose money because it begins as can you get access to low cost power? It then continues to Once you have access, can you get permitting? Does that take long or do you have real access that gets you fast? Permitting. Once it becomes a construction project, can you keep control of your costs once it's finished, can you keep good tenants in it? The ways to lose money in property are large and many. There's no free lunch there either. When you are trying to go unbelievably quickly, it's harder and harder to be disciplined and not make mistakes.

[Harry Stebbings]

To what extent is it important to be fully horizontal? You know, we hear about Zuck wanting the data center build out to be just immense in terms of size and scale. To what extent does it need to be horizontal versus vertical?

[Andrew Feldman]

It is completely unclear. The most successful two companies to date, OpenAI and Anthropic, neither are vertical. OpenAI used Azure 100% infrastructure for years and Anthropic has used a combination of AWS and Google. And so neither are vertically integrated to date. Now whether that's the right strategy going forward, whether they'd make those decisions again, but it's clear that it's not the only strategy that there are plenty of working models where you are not fully integrated from chip through system, through data center through software, all the way to the top.

[Harry Stebbings]

Again, sorry to cite it, but it's kind of handy having just done it. Jonathan said that you would definitely have OpenAI and anthropic build out their own ships because then they would have control of their own destiny. Do you think OpenAI and Anthropic build their own ships so they don't have self reliance on Nvidia in the way that they do today?

[Andrew Feldman]

I think that there is a long history of software companies failing to build chips. The list is very large. Whether OpenAI can do it, whether they can do it through partnership with other vendors, with Broadcom, with smaller more innovative companies is an open question. Companies at the size of Microsoft have been unable to deliver chips. There are plenty of examples as you look across the FAANG group where chips were tried. I mean probably the most successful is Google and they're 10 years in. Modern software does not fit well in a chip making framework. Weekly sprints don't work well on two year long projects. Move fast, break things often is not the way you think in the chip world. The way you think in the chip world is measure twice before you cut once because your bugs cost you six months and tens of millions of dollars. It's a very different mentality and where there's been success it has frequently been acquired. Apple got into the chip business through buying psme. Amazon got into the chip business through acquiring Annapurna. Google acquired the talent from a collection of companies and then set it in a BU that was aside and under somebody who had enormous respect in the organization in ORS and had a 10 or 15 year view. These are things that have been challenging in in many companies. Chip building is an MBA nightmare. Your analysis that says intel had, you know, between 2000 and 2010 some of the world's leading architects, the world's leading fabs, and proved completely unable to build a working cell phone part. And you ask yourself why they had everything they needed. And you know, you do an MBA chart and it's like it's impenetrable. And the answer is this is really hard. Very small sort of mental model differences produce tremendously different results. How did every leader miss the largest compute market in the first part of the 21st century? How did AMD miss it? How did ARM win it? All the leaders missed it. And then you say all right, maybe there's something in the guts here that I don't understand. Right. You got to really get in there. And it's not on a PowerPoint, it's not in a two by two. It's not at some sort of consultant level. It is deep in the DNA of the small number of people who can build these things. You know we are lucky at Cerebras we've got one of the top six or eight teams in the world. Other startups don't.

[Harry Stebbings]

What does that market look like do you think in 10 years time? I know 10 years is a huge amount of time given where we're at, but in 10 years time is it a monopoly market with one taking 90%? Is it like cloud where.

[Andrew Feldman]

Which part of the chipmunk we talk about AI Silicon or we're talking about silicon in general or we're talking about.

[Harry Stebbings]

I would say silicon in general.

[Andrew Feldman]

Yeah, absolutely not one takes 90%. You know, even at the at intel strength they had dominance in x86 and 0 market share on the cell phone and almost no share in the switching market. Broadcom had dominance in the switching silicon market which is a form of processor and silicon and no share in x86 or other forms of compute. Yeah, it will not all accrue to one or two companies.

[Harry Stebbings]

Andrea, how do you think about the importance of margin today as a business? Is cerebrous?

[Andrew Feldman]

Well, given I think the reason we were able to raise at a higher valuation and from better investors and more money is because we had them and others who were out looking for mark looking for money had negative margins. As you prepare for being a credible public company, people do look at your margins. That's a really important part of moving from being an idea to being a real company.

[Harry Stebbings]

What are Nvidia's margins today?

[Andrew Feldman]

Extraordinary. Some of the highest in history for a hardware company.

[Harry Stebbings]

How do you think about that? Is that just pricing power which they are taking advantage of?

[Andrew Feldman]

Absolutely. I mean the short answer is why does it make sense for AWS to build a trainium part? Well, because they want to get rid of the 78% gross margin that Nvidia is charging them. It might be on the high end chips, 85% people don't like that. Historically. Historically people sort of put that in the back of the mind and they remember it. When intel stumbled, the number of people who came out of the woodwork to kick them when they were down was extraordinary. And it was sort of years of pent up frustration when the giant stumbles. We've seen that again and again.

[Harry Stebbings]

Speaking of giant stumbling and being built, do you think sovereignty will be a big enough reason why incumbents are built? We have, like, Mistral, a model provider in Europe, and kind of sovereignty is their core play. Do you believe that is a sufficient enough core play to be a giant?

[Andrew Feldman]

Well, I think right now, sovereignty plus the fact that we deliver their inference through the fastest hardware on Earth makes their product, the lechat product, really compelling. They're using their advantages to compete. There were in Europe, too few AI labs that are doing interesting work. They sort of looked around and sort of used strategic advantage. You know, we want to be the Europe's leader. You know, played that card really well. Hats off to them. And then they raised, at a huge.

[Harry Stebbings]

Valuation, a final one, just in terms of geography. Deepseek obviously had that moment, and it kind of solidified the concerns around China. How do you feel about China today as a pressing concern towards the US in terms of the race towards AGI between the two? Do you hate the way that it's posited as China versus the us, The AI race? How do you feel about that?

[Andrew Feldman]

I think it benefits neither the position we're in. The arms race certainly didn't help either the US or Russia in the 80s and 90s. We both spent money we wish on weapons we wish would have been spent on infrastructure, people, or other things. We will be much stronger if we can find ways to peacefully engage. Before these issues, we knew the guys at didi and bytedance and Ali and Baidu extremely well. They're talented engineers trying to build cool stuff. I think our governments are at loggerheads, and that's a problem. We had a huge opportunity in 2019 to do a deal in China, and I decided to pass because I didn't think it was the right thing to do. Long before the Department of Commerce limited exports to China, I didn't think it was right, and I was concerned about how the technology would be used. The realpolitik right now is that they're better at making drones, they're better at making robots. Their government has an extraordinarily aggressive policy in AI. For years, they backstopped their venture groups. So if you lost money in an AI company, the government would make you whole. Imagine that, Harry. Imagine how much money you could make if the government of the UK offset some of your losses from AI companies that didn't work out. You know, we have real work to do in the us.

[Harry Stebbings]

What work do you have to do that you haven't done? What would you like to see?

[Andrew Feldman]

I think China thought long and hard about Their power infrastructure, their form of government allowed them to plan strategic. Our decentralized form of government has left us with sort of a patchwork of power infrastructures. Even if the federal government wants to support you, there are local regulations, like at the city and county level of towns that can interfere with a project and set a project back billions of dollars. I mean, Samsung built a FAB in Texas and they had to change the design of a fab because a local fire ordinance in the US government worked for years to get deployment of billions of dollars in Texas. And a local fire ordinance set them back eight months, 10 months and caused them to redesign the fab. That's a problem that we have to sort of collectively work through. We have the premier universities, historically drawn talent from around the world. If you look at the great CEOs in our industry, Janssen Hock Tan, Lisa, I mean, you go down the list at Sundar, at Microsoft Soft, they came and their parents came. We got to take that really seriously.

[Harry Stebbings]

You don't buy the whole, well, actually a load of people actually just abused H1s and we'll just move to O1s, which people were using anyway. And the average salary for an H1 was $120,000. It's a good thing. And people would just use O1s.

[Andrew Feldman]

I am sure that in every government program there's abuse. Was there more abuse in the H1 than in other areas? I don't think so. Having the best and the brightest come to your universities and once they benefit from sort of our great institutions, to want to stay and contribute first with a J1, which is the student visa, and then enter the H1B lottery through the approved process to get a green card and become citizens. And this is how my parents did it. Right. I think it's one way to bring an extraordinary amount of talented people to.

[Harry Stebbings]

The U.S. is there anything else you changed? You said the power infrastructure and the permitting around it.

[Andrew Feldman]

Power infrastructure ends up at the local level, which is not necessarily where big ideas and sort of strategy is well knitted together. We have starved our universities of compute. If you want to do interesting training work at a university, very hard to get enough compute to do that. We're just not set up for that. Those are two dimensions. I think the Trump administration's done a good job in generally relaxing some of the regulations that were painful.

[Harry Stebbings]

Andrew, I want to do a quick fire with you. So I'm going to pummel you with quick questions and you got to give me your immediate thoughts.

[Andrew Feldman]

That's hard Because I only have long answers, Harry.

[Harry Stebbings]

So what do you believe that most around you disbelieve we will have peace.

[Andrew Feldman]

In the Middle east in our lifetimes.

[Harry Stebbings]

Why do you believe that?

[Andrew Feldman]

Having visited and spent time now in the uae, in Saudi and Qatar, I think the returns to moderation, the economic gains, someone said we're too busy to hate right now. We're too busy building. Those have been writ large so clearly in the uae. With the rise of Dubai and the UAE in return for making peace with Israel, in return for a more moderate position, I really believe that that is the path to the future.

[Harry Stebbings]

How much of your revenues are from the UAE?

[Andrew Feldman]

I think in the S1, it says maybe 1H24. We haven't published others, but a lot. 75, 80%.

[Harry Stebbings]

I mean this in the nicest way. Do you not have to say nice things about it then? If someone's giving me your question.

[Andrew Feldman]

No. I went there to do business as a Jewish guy before we had any business done. Right. What I found surprised me. And we don't do much in Saudi, and I think they're making great strides. And we don't do anything in Qatar right now. I think they're making great strides. It may well sort of be colored by the fact that I spend time in Abu Dhabi and I spend time in Dubai and I spend time in Riyadh, and I spend time in Doha. Sure, it's colored by those things, but.

[Harry Stebbings]

I think, why are your revenues concentrated there? Is it just because they're, like, more willing to embrace innovation, new relationships, new vendors?

[Andrew Feldman]

They bought so much, they consumed. And, you know, the data I gave you was through 1H24, they placed such big orders. They consumed our manufacturing capacity. They are building at such an extraordinary rate that through 1H24, they consumed an enormous amount of our manufacturing capacity.

[Harry Stebbings]

Did their orders exceed your expectations of their orders?

[Andrew Feldman]

I think their orders exceeded everybody's expectations. You can be a professional salesperson in Silicon Valley for 20 or 30 years and not see a $500 million order. You can go around the Valley right now and talk to VPs of sales or EVPs of sales at dozens of public companies who've never seen an order of that size. They were bold and they were early. And, you know, when we started doing business with G42, nobody heard of them. Not everybody in the world heard of them.

[Harry Stebbings]

Do you think that was a resource planning mistake from you? I mean that nicely.

[Andrew Feldman]

They're all mistakes. In retrospect, if we hadn't won Them and we had the resources for it. That would have been a resource planning mistake. I mean I'm, I'm in the business of, of making big bets and making lots of mistakes.

[Harry Stebbings]

Harry, what was the biggest bet you've made with Cerebras that didn't work out?

[Andrew Feldman]

My bets here have, have been pretty good to go to wafer scale solve a problem that nobody had previously solved. Gene Amdahl, one of the fathers of our field, failed. IBM failed. TI failed. Everybody's failed at this. We had a period of about 15 months between about 2017 and early 2019 where we couldn't make one. And we were running a burn of about 6 million a month, 7 million a month. And we stayed with it and our board stayed with it. The result was do you have signs.

[Harry Stebbings]

That it would work?

[Andrew Feldman]

Yeah, we did. And we weren't running around like chickens without our heads. We were to going, going through engineering process. Each failure was, you know, we did a full fa, a failure analysis each time we fixed the cause, we did another one didn't work. Did another one didn't work. And each time we got a little better and we got better and better and better. Then we solved it. The first one that worked, the founders were in a tiny little lab that was a converted conference room that for cooling we had the windows open and we'd blown a hole in the wall so we could get external a chiller outside and poured in. And when we had it running, the founders stood there together and stared at the box running, which is about as interesting as watching paint dry. And we stood there and we couldn't believe it. It was like we have just solved a problem that for 75 years the smartest people in our industry have been unable to solve. And we stood there for like half an hour and it was one of the highlights of my career.

[Harry Stebbings]

Pretty cool. All right, I'll give it to you.

[Andrew Feldman]

So that was a big, big bet. That was fair enough.

[Harry Stebbings]

The 6, 7 million a month burn. I'm like, all right, fair. I'm almost picturing like angels singing in this, tears coming down your face.

[Andrew Feldman]

You know what? It felt like that. And it was the brainchild of my co founders. It was their invention. It was a physical manifestation of their ideas.

[Harry Stebbings]

Where are people investing today where they will completely lose their shirt.

[Andrew Feldman]

The silicon industry is not place for 25 year old CEOs. No matter how smart you are. The returns to having built parts before in what we do are enormous. The number of different relationships that are necessary. You need a relationship with the fab need a relationship with the EDA toolmaker, you need back end design engineers, you need logic design engineers, you need IP relationships with IP providers. It has been an extremely difficult road for young CEOs. On the other hand, young CEOs in many of the markets you invest in have done extraordinarily well, particularly where they look like their customer. The reason that the entire social networking world was built by young founders is they were building a product for their friends and that is an advantage. The reason that AI startups who are doing tools for other students for coders are young is because they understand the needs and demands of their target customer base extraordinarily well. That's an area where people are going to get clobbered is to take a mentality. That said it's enough to be smart in this field to build a good chip. That has historically not been the case that there are real returns to having done 15 or 20 of these in the past.

[Harry Stebbings]

Where are people not investing enough where they should be investing more?

[Andrew Feldman]

There are this collection of extremely unsexy things that are causing tremendous pain across the industry. Data cleaning, your data pipeline. These things are. Nobody puts on their LinkedIn data pipeline expert and yet these are some extraordinarily valuable cats. Nobody leads with a leader in the cleaning and tokenization of data. And these are extraordinarily important roles. Many AI projects fail on those fronts have nothing to do with the AI. They fail because the data was a disaster. They fail because everything except the AI was a failure. That's an area that profoundly under invested in.

[Harry Stebbings]

What do you think the data provision market looks like? We see surge, macaw, invisible, Turing handshake, moving into it more and more. What does that market look like in five years? All of them are above 100 million. ARR.

What the fuck happens there Andrew?

[Andrew Feldman]

I don't know. That is a very curious market.

[Harry Stebbings]

Why is it curious?

[Andrew Feldman]

Well, scale sort of pioneered it. Turing was in a different market completely and pivoted to it and found great success in it. There are these collections of others. I think clearly the provisioning of value added, tagged or evaluated data is really important. Whether it's durable, whether we get machines that do it every bit as well as people is a question that's really hard to answer right now. That's why it's curious is that clearly it's important now and the question is will it clearly be important in three years? That's a question I don't know the answer to. Maybe it could go either way.

[Harry Stebbings]

What's Your craziest prediction in terms of how AI reshapes the future. In five years, for example, Jonathan said, hey, I think AI will create massive labor shortages. It will create so many jobs for so many people that we will have.

Massive labor shortages in five years.

[Andrew Feldman]

Absolutely wrong. I think economic dislocation isn't resolved in very short periods of time. That might be true in 15 years, but in the three to five year time frame, I certainly don't believe that will be the case. The adoption of AI or the diffusion of AI into the economy, it will nibble its way in. Let's ask this question. Alphafold solved one of the hardest problems in chemistry, a problem that had been open for years. Name a drug that's resulted from it. Not one. Now, I believe there will be, but AlphaFold's four years old now. This was a massive breakthrough for which the inventors were given Nobel Prizes. Where's the drug? Show me the medical benefits. It will get there. It will be important. Continuations of the model will have fundamental impact. But where are the X ray crystallographers who are displaced because of it? That was what X ray crystallographers were doing, only physically. They're not out of work. In fact, there's more demand for them. I think it will have really interesting effects on the way we educate children. And that's an interest of mine. We've sort of been educating children the same way since Alexander the Great was tutored by Aristotle, right? And it was sort of like, get a smart person. They're older, they stand behind you, they tell you what to do, read, you read it, you talk to them about it, they correct your paper. This form of instruction has been sort of unchanged. Maybe YouTube changed it a little bit in that you had different instructors. But imagine a system where, for example, you made a set of mistakes in your math work. The result wasn't read on a paper. Oh, look, you got it wrong here. But they compared the type of mistakes you made to the type of mistakes thousands of other students students made and said that for this group of mistakes, we have found that the following workbook is extremely effective at remedying this hole in their thinking. Imagine that nobody does it. Nobody differentiates and modifies the training based on the type of error the students are making. And that's exactly what you ought to do. So I think the way we teach will change a great deal what it means to be entry level and a company will change a great deal, because what entry level has generally meant in consulting firms at investment banks has been doing Shit work. In particular, being really good at spreadsheets and writing summaries of other people's research. AI will be better at that. That will change a great deal. I mean, I always thought that was a terrible way to spend an extraordinary. You're 22 through 24 years years you are coming out of top schools. There's so much to be learned, there's so much you can contribute. But to do huge hours doing spreadsheets, I think there's vastly more productive thinking those students that those young people are capable of, and more learning that they can do and therefore be vastly more productive in the following years. And I think AI will change that.

[Harry Stebbings]

Final one for you. Andre, what would you do if you knew you wouldn't fail or couldn't fail?

[Andrew Feldman]

I guess I don't. I've never thought of that. I look at it the other way, is that every day I go to battle with Goliath. Every day, every dollar we sell is a dollar that if we didn't work at it, if we didn't think, if we didn't invent, if we weren't 10 times better, would default to Nvidia. And before I competed with Nvidia, I competed with Cisco. 15 years. And every dollar that we sold there, if we didn't build better product, if we weren't more aggressive, if we weren't more creative, would have defaulted to the market share leader. And I think for me in my career, I take great pride in facing every day the most wicked curve, the curveball pitcher for your cricket example, you know, the scariest spin bowler, the scariest speed bowler. I enjoy that. That is something that in a quiet moment, you sit back and say that I'm competing with every disadvantage against the. The absolute best in the world every, every single day at work. And I love that. And what's more, everybody's betting against me except a very small group of people who you named, who stood up early on and said, you know, maybe he can beat them. That's the life I've chosen and the career I love.

[Harry Stebbings]

That is such a good end as well.

You know, I do obviously many shows, there are some ends where we can't end like that.

That is like a depressing end. That's a fantastic end.

I so appreciate that.

I so appreciate you. You know, you're my go to when I'm trying to understand what the fuck's going on. And so thank you so much for explaining this to me today, dude.

[Andrew Feldman]

Well, I'm happy to jump on my view of your LinkedIn post, by the way, is there are two exactly two things I've read in my life that feel like they understand what entrepreneurship is. And the first is Ben Horowitz's book the Hard Thing About Hard Things. And the other is your Tweets and your LinkedIn posts. They have the feel of what my life is. This notion that somehow you can achieve greatness. You can build something extraordinary by working 38 hours a week and having work life balance. That is mind boggling to me. It's not true in any part of life. And your willingness to jump in and say, no, that's not how it's done, guys. You can have a great life, you can do many really good things and there are lots of paths to happiness. But the path to build something new out of nothing and make it great isn't part time work. It isn't 30, 40, 50 hours a week. It's every waking minute. And of course there are costs. It's probably true for world class athletes too if you listen to what Ronaldo talks about. I mean he worries about everything he puts in his body. He trains every single day. They work on rest, right? Rest isn't rest. Rest is something you work on so your body rejuvenates faster. This isn't 30 hours a week or 40 hours. These guys are the best in the world at everything.

[Harry Stebbings]

But before we dive into the show today, I love seeing the team come together to make this show happen. What I don't love is trying to keep track of all the information, the data and the projects that we're working on across dozens of platforms, products and tools. That's why why we use Coda, the All In One collaborative workspace that's helped 50,000 teams all over the world get on the same page. Offering the flexibility of docs with the structure of spreadsheets, Coda facilitates deeper teamwork and quicker creativity. And their turnkey AI solution. The intelligence of Coda Brain is a game changer. Powered by Grammarly, Coda is entering a new phase of innovation and expansion, aiming to redefine productivity for the AI era. Whether you're a startup looking to organize the chaos while staying nimble, or an enterprise organization looking for better alignment, Coda matches your working style. Its seamless workspace connects to hundreds of your favorite tools including Salesforce, Jira, Asana and Figma, helping your teams transform their rituals and do more faster. Head over to Coda iO20VC right now and get six months off the team. Plan for startups for free. That's Coda C O D A IO 20 VC and get six months off the team plan for free. Coda IO 20 VC and talking about trust Today customers expect it faster than ever and that's why over 10,000 global companies trust Vanta. Vanta automates up to 90% the of of the work for in demand compliance standards like SOP2, ISO 27001 and more, using smart AI to centralize workflows, manage risk and get you audit ready in weeks, not months so you can stop chasing paperwork and start closing deals. And a new IDC report found that Vanta customers achieve $535,000 per year in benefits. That's insane. And the platform pays for itself in three months. I had no idea about these. Whether you're growing fast or just getting started, Vanta connects you with trusted auditors and experts support to help you build trust with customers. Get $1,000 off your first year at Vanta.com 20VC that's Vanta.com 20VC.

You have now arrived at your destination.