Google and Meta CEOs both out in last 24 hours now agreeing with my AI Arms Race narrative.
There are 4 major players in this game theory problem, and two are now on-the-record about their strategies.
19
91
559
122,031
Since starting at @sequoia, I've been reflecting on the massive CapEx buildout in AI.
For today's GPUs to payback, $200B of revenue will need to be generated per year of CapEx.
Long-term, this is good for startups. Short-term, it could get messy.
bit.ly/3RotkBH
22
54
309
156,582
With Character dropping out, the qualifying round of the AI race is now officially over.
Congrats to the finalists: Microsoft/OpenAI, Amazon/Microsoft, Google, Meta and xAI.
The next phase begins: One defined more by the construction worker than the research scientist.
20
2
201
59,406
NVDA was asked our $600B question on Q2 call: Where is the customer's customer's revenue?
Their answer:
1. Traditional workloads will move from CPU to GPU
2. ChatGPT and coding AI
3. Meta saves $ using GPUs for algos
4. Countries buying GPUs
Is this enough to justify the hype?
9
10
154
32,818
Bootstrapped to $1M ARR with 4 people. Now: $10M ARR, 2,500+ customers.
Juicebox is more than a tool, it’s a new way of recruiting. Proud to back @davepaffenholz and @ishangpta—their grit & hustle is next-level.
Excited to announce Juicebox has raised $36M in funding, including our Series A led by Sequoia Capital (@sequoia) - to help the world’s leading teams win the talent war.
More below 🧵
7
6
155
32,457
Nvidia just became the most valuable company in the world.
In September, I published AI’s $200B Question.
The question I had then was: “Where is all the AI revenue?
Below is an update on what’s changed & what hasn’t:
sequoiacap.com/article/ais-6…
15
22
148
102,132
AI labs are starting to look like pro sports teams.
Backed by deep-pocketed owners.
Star players making 8-figure packages.
Fierce, fluid battles for top talent
New “draft class" each year
Why talent — not compute — is now the AI bottleneck:
sequoiacap.com/article/why-a…
11
10
145
21,975
Morning after last night's "AI Super Bowl."
Here's where we stand on AI's $600B question:
(1 & 2) Nvidia: CPU to GPU
(3 & 4) Meta: AI game theory
(5 & 6) Google: Underinvesting vs. overinvesting
(7 & 8) Microsoft: M365 & Github copilots
(9 & 10) Amazon: Data center logistics
7
18
131
29,872
Excited to share our launch of Kela Defense from stealth today, including a great profile in the @WSJ.
As my first Sequoia investment, this company has been a huge focus for me over the last year. We’re excited to finally share what we’ve been working on.
5
10
125
17,659
New post: AI CapEx Now Hinges on Deus ex machina
With talk of 100 GW or 250 GW of energy buildout, a really big plot twist would be needed to make the math pencil out.
Nothing short of AGI will be enough to justify the investments now being proposed for the coming decade.
12
3
102
26,784
Here’s the question now being asked all across the AI ecosystem: Is there a way for someone else to take on the demand risk from AI, while I capture the profits? (1/3)
8
16
96
24,945
New Post: Circular Deals & Supply Chain Dynamics
1/ The big story of 2025 has been the direct transfer of data center demand risk from hyperscalers to NeoClouds and chip companies ("the demand hot potato"). Circular deals crystallize this dynamic.
7
17
98
36,571
The next big catalyst in AI is the launch of 100k clusters. Everyone has committed to this path.
If scaling laws hold at 100k, we'll get another leg of CapEx. If not, things will slow down.
Given physical requirements, this is ~1+ years away. We may be in limbo for a while.
11
6
84
23,200
Big day in the market. Zuck & Sundar acknowledged how much AI CapEx is FOMO vs. actual revenue. Public market grappling with updating their AI narrative.
Full breakdown here: sequoiacap.com/article/ai-op…
2
9
80
16,010
New post: What does it take to win in the next phase of AI? Our answer: Steel, servers & power.
Today, 5 companies have qualified for AI's next race: The race for data center scale-up.
The market structure has now crystallized and the gritty execution phase begins.
3
4
67
13,453
New blog post:
1/ America is in the "Age of Bubbles." From Dot-com to 2008 to today, we’ve been riding a wave of speculative booms. AI is the latest. But what happens when the bubble economy finally pops?
7
8
70
14,795
Thanks for having me back @HarryStebbings. Really enjoyed the conversation, which ended up being quite timely. We coined "Steel, Servers & Power" last time. Will be interesting to see what stands the test of time from this one.
Our last episode with David Cahn was one of the most downloaded of last year.
This episode today is even better.
- The Winners & Losers in AI
- The $0-$100M Club: Is Triple, triple, double, double dead?
- The Future of Defence: Who Wins & Who Loses
- How to Analyze Margins & Growth Rates in the World of AI.
Spotify 👉 open.spotify.com/episode/6Cx…
Youtube 👉 piped.video/aGoUu5VLvLY
Apple Podcasts 👉 podcasts.apple.com/us/podcas…
Timestamps:
00:00 Intro
01:09 Why Building Physical Data Centres is a Moat
10:36 Are We In an AI Bubble?
14:42 Winners and Losers in a World of AI
16:09 The Role of Big Tech and Monopolies
22:22 Breaking Down Circular Deals in AI: The Truth No One Sees?
34:26 Why Kingmaking is BS and VCs Do Not Make or Break Companies
37:53 The Importance of Margins in AI Investments
39:54 The $0-$100M Revenue Club: Is Triple, Triple, Double, Double Dead?
49:13 Why the Most Important Hire for Startups Today is 23 Year Olds
58:29 The Future of Defence: Who Wins and Who Loses
01:07:33 Quick-Fire Round
My 6 takeaways with @DavidCahn6👇
2
5
67
29,413
AGI isn’t about machines outsmarting everyone overnight.
It’s about raising the floor.
Introducing the 5% Rule: whenever AI beats the bottom 5% in a field, billions in value are unlocked.
We’re already there in writing, coding, customer service, therapy & companionship.
10
5
57
10,514
Every time a new AI interface comes out, I try it. I'm convinced that the right conversational interface for AI will have tremendous demand and will change the way we interact with the digital world.
6
5
60
15,785
Wonderful thread from @gradypb on the next generation of @sequoia. What makes Sequoia great is the depth of the team, the rigor of our collaboration, and the relentless effort to leave the firm better than we found it.
Steel sharpens steel. It’s a privilege to be pushed and sharpened by this team.
1/ We @sequoia have always said that we’re only as good as our next investment.
It’s also true that we’re only as good as our next generation
@Alfred_Lin and I get to work with the best generation yet:
1
2
59
19,005
It has been a true honor to study under one of the all time legends of the investment business, @roelofbotha.
Excited for the next generation of stewards with @gradypb and @Alfred_Lin. Sequoia's track record of generational transition is unmatched.
1
3
51
6,471
The ante to enter this next race is a minimum of five billion dollars. Enter the era of steel, servers, and power.
sequoiacap.com/article/steel…
8
5
50
5,184
Silicon Valley is obsessed with AGI.
But AGI is not the only path to a better future.
Bringing 1B developers into the software economy could rival AGI in terms of its economic impact – and it’s already happening.
sequoiacap.com/article/the-n…
10
16
46
13,572
Thanks @HarryStebbings for a great discussion & deep dive on AI's $600B question:
👉 The revenue hole that needs to get filled
👉 The AI arms race and cloud oligopoly
👉 Why overbuilding is good for startups
👉 Steel, servers & power
👉 The need for a cash machine to fund AI
I’ve been fortunate enough to do AI shows with @Sama, @ylecun & @arthurmensch.
But this show today with @DavidCahn6 at @sequoia is the best of all of them:
🧠 AI’s $600BN question
🏃 Why no one will never train a frontier model on the same data center twice
🏗️ Servers, steel & power: The 3 pillars of the future of AI
Not one to be missed, my 8 key takeaways 👇
5
6
50
17,631
Imagine you knew for certain that AI was going to be as transformational as the internet, and that you control the only AI company in the world. How fast would you build CapEx? (1/11)
3
6
48
39,977
For more than a decade, @amasad has been working to unleash the next 1B software creators. With the rise of “vibe coding” and the growth of @Replit's user base to tens of millions of users, this vision is materializing.
Thanks to @sonyatweetybird for having us on Training Data.
2
3
44
13,014
Joining @tbpn today to talk about the AI talent race.
Since our blog dropped Tuesday, news broke on Meta's $100M+ signing bonuses and there are fresh rumors of an NFDG acquisition.
The pace has gone from fast to breakneck.
Morning. Here are our guest call-ins today:
- @mikeknoop (ARC Prize)
- @DavidCahn6 (Sequoia)
- @walden_yan (Cognition)
- @eoghan (Intercom)
- @jeff_weinstein (Stripe)
- @GarrettLord (Handshake)
- @tanay_tandon (Commure)
See you all on the stream.
2
4
40
8,854
Original post laying out game theory dynamics in AI CapEx race: sequoiacap.com/article/ai-op…
6
34
5,639
I’m not convinced by the game theory of Stargate:
- OpenAI wants maximum CapEx
- Microsoft wants optimal CapEx (ROI matters for them)
This will always be a natural tension.
Outsourcing CapEx to Stargate doesn’t resolve this. It only moves the financial burden.
Good chess move or bad one?
9
34
9,919
Last January, we compared ChatGPT to AI’s “Big Bang” and predicted 2024 would be AI's "primordial soup" year.
“There is much potential in the air, and yet it is amorphous," we wrote at the time.
Now, the AI ecosystem has hardened. In a new post, we share 3 new predictions:
2
4
32
11,339
Broader deep dive here on what comes next in the AI CapEx race, and specifically on the path to building 100k and 300k GPU clusters.
sequoiacap.com/article/steel…
4
6
30
3,694
Next post in the 200b / 600b series turns from the revenue side of the equation to the cost side. We look at the data center construction boom and make 5 predictions for how this will affect the energy sector and the economy more broadly. (1/7)
3
6
29
13,598
ChatGPT was AI's "Big Bang" in late 2022.
2023 was a frenzy. There was a view that all the hard problems had been solved, and AGI is inevitable.
2024 will be about facing down ambiguity and seeking out hard problems. This will unleash AI's impact.
sequoiacap.com/article/ai-in…
5
6
29
13,696
AMD announcement today of the ZT acquisition reflects new reality in the AI supply chain.
Now that the competitive landscape has hardened, everyone is looking at profit margins & demand risk.
ZT improves both margin capture and customer acquisition.
sequoiacap.com/article/the-a…
4
4
29
8,587
For the last year, people have been thinking "Google is spending $XB on CapEx, they must know something we don't know about the future of AI."
What Sundar confirmed yesterday is they don't know anything the public doesn't already know -- they just have different incentives.
2
2
24
2,490
Thanks to @l2k for hosting me on the Gradient Dissent podcast.
We start on the recent Stargate and DeepSeek news, revisit the $600B question, and go deep on topics like how religion and AI intersect.
🎙️ DeepSeek, Stargate, and AI’s $600 Billion Question
On this episode of Gradient Dissent, @l2k sits down with @sequoia’s @DavidCahn6 for a discussion on current AI investments and its evolving landscape.
What does this mean for AI spending and growth strategies?
Tune in! 👇
4
6
26
16,057
Thanks to @HarryStebbings for having me back on the podcast for a sequel to our "Steel, Servers & Power" episode last year.
In this episode, we turn to the AI bubble narrative, which has gone from contrarian one year ago to consensus today, and discuss who will win long-term.
2
3
30
8,590
Great to interview @chaselochmiller on Friday at #AIAscent about the AI industrial revolution.
I've been following the "Steel, Servers, and Power" world closely, and Chase is at the absolute center. I learn something new every time we speak.
bit.ly/4d9KYSA
2
3
26
15,499
Thousands of recruiters, including our own team, are already obsessed with @davepaffenholz and @ishangpta’s Juicebox. We’re excited to be leading their Series A.
techcrunch.com/2025/09/25/ju…
1
24
8,482
In part one of this piece, we’ll walk through the tug of war between supply chain players over risk and profit. In part two, we’ll unpack the instability of today’s equilibrium. (3/3) sequoiacap.com/article/the-a…
2
21
2,010
At Fully Connected, I talked with @LavanyaShu97869 about building AI startups: how to recruit top talent when Big Tech is offering massive packages, why mission still wins, and what founders should focus on right now.
Thanks for having me
piped.video/DmfVlf1yHb4
3
4
19
17,629
Link to post: sequoiacap.com/article/ai-in…
2
17
6,211
In the context of AGI, Deus ex machina now has a double meaning: the literal creation of a new godlike intelligence, but also the only path forward to resolve an increasingly tricky ROI problem.
1
1
14
1,730
5/ All of these deals are quoted in GW, which obfuscates just how massive they are. And how much money is still needed to a) Pay for them, and then b) Generate an ROI. Below is a table that converts from GW to $:
2
2
17
1,700
There are a few reasons for this, but the biggest is that PhD programs are being starved for talent, and it's very hard to make true low probability, divergent research bets. But this type of divergent search is required for scientific progress.
3
14
1,140
It is critical that we stand with Israel during this important time. #StandsWithIsrael
It has been alarming and heartbreaking to watch the recent events unfold in Israel. The terrorist attacks by Hamas on innocent civilians are reprehensible. We have been in touch with our founders, former partners, and companies with operations in the region to see how we can best support them. Our hearts go out to those affected during this horrific time.
15
3,261
What makes a great founder?
This framework explores the founder journey -- starting with hardcore science & ambition, and breaking down how that can evolve into invention & visionary thinking with the power of intuition.
bit.ly/3U1vRRZ
2
15
1,991
6/ A lot of people have been talking about how the AI CapEx buildout is debt financed, but in reality, what’s interesting about the circular deals is precisely that they are not debt financed. For now, debt investors are unwilling to bet on data center demand 15-years out
1
2
14
2,276
Prediction 1: LLM providers have evolved distinct superpowers — this should lead to incremental differentiation and a contested pecking order in 2025.
· OpenAI = Brand
· Anthropic = Talent
· xAI = Data Center Scaling
· Google = Vertical Integration
· Meta = Open-Source
2
14
758
Today, Big Tech companies have stepped up to alleviate some of this tension. They are acting as risk-absorbers within the system, taking on as much demand risk as they possibly can, and driving the supply chain toward greater and greater CapEx escalation. (2/3)
2
13
1,993
4/ I’ve been thinking and reading about these ideas for a while. With the DeepSeek correction yesterday, now feels like a good time to share:
dcahn.substack.com/p/its-all…
2
1
13
2,304
This is surprising, since AI luminaries have been walking back their AGI timelines. If anything, overinvestment in compute may reduce the probability of reaching AGI over a longer time horizon (ten to twenty years).
2
13
2,127
Great interview with @roelofbotha on how to navigate AI and macro + @sequoia's continued investments in European startup ecosystem
Our message to founders today? Focus on the fundamentals.
@roelofbotha spoke with @emilychangtv about thinking long-term, building in a challenging market, our enthusiasm about the European startup ecosystems, AI investments, and more.
1
10
2,657
Prediction 2: AI Search is emerging as a killer app—in 2025, it will proliferate.
· Sematic search with AI is more powerful than navigational search
· Perplexity has captured lightning in a bottle
· Specialized search engines for doctors, lawyers, investors, etc. will emerge
1
1
12
873
h/t @amasad @kiwicopple @AntWilson who have greatly influenced my thinking on this topic.
@amasad was one of the first to articulate the 1B goal and @kiwicopple @AntWilson had the vision and courage to envision a world where developers can build in a weekend & scale to millions.
1
11
1,517
7/ Full post here: dcahn.substack.com/p/circula…
2
2
12
2,098
With CapEx plans now firmly in place and the competitive landscape set, the new AI era begins. In this new phase of AI, steel, servers and power will replace models, compute and data as the “must-wins” for anyone hoping to pull ahead.
1
10
3,539
Congrats to the W&B team @l2k @shawnup @vanpelt @lavanyaai!
Amazing to see the journey from <$1m ARR six years ago to more than 1M users and a pivotal place in the AI ecosystem in 2025.
Today we announced that we are being acquired by @CoreWeave, the AI Hyperscaler. 🪄🐝
We could not be prouder or more excited to join forces with this team.
Our CEO, @l2k, wrote a blog post with more details:
wandb.ai/wandb/wb-announceme…
1
10
2,376
4/ But these companies, even combined, are much smaller than Microsoft and Amazon, and so now Nvidia, AMD and Broadcom are stepping up to absorb some demand risk, too. There is also a sense in which these deals are “priming the pump” to attract further investment.
1
2
10
1,465
Cloud competition and game theory is turbocharged by supply constraints. If you don’t acquire land, power and labor now, someone else will (10/11)
2
1
10
2,318
Many market participants today would have you believe that there is a choice between being an “AI bull” and an “AI bear.” This is a false dichotomy. The CapEx debate is a debate about speed, not about magnitude (3/11)
1
9
1,208
Prediction 3: ROI will remain problematic and CapEx will begin to stabilize in 2025.
· AI's $200B question and $600B question remain unanswered
· In 2024, Big Tech was nervous about falling behind competitors
· Now, they have their arms locked firmly around the AI revolution
1
10
2,044
Up until now, you could fit your training cluster into an existing data center via colocation or retrofit.
This is changing: The next generation of models are aiming for 300k GPUs clusters. To house one of these models, you need to build an entire new data center.
1
9
881
Deus ex machina (“god from the machine”) refers to a plot device where an unsolvable problem is suddenly resolved by an unexpected intervention...
1
9
930
No matter what you believe, all of this building will be fantastic for startups. Infrastructure providers are bearing most of the risk in AI. This is a subsidy for startups building on top of them. With all this investment, training & inference costs will keep coming down (8/11)
1
10
1,229
Launching a Substack for those who want to get future AI posts via inbox. You can subscribe here:
dcahn.substack.com/
1
10
1,357
Full post here: dcahn.substack.com/p/ai-cape…
2
1
10
1,540
Five companies have arrived at the starting line in this new race toward data center scale-up: Microsoft/OpenAI, Amazon/Anthropic, Google, Meta and xAI. Each has a model that has held up against serious benchmarks, and the necessary capital to proceed.
1
8
785
The race to model parity has been the defining project of the last 12 months in AI.
The next phase in the AI race is going to look different: It will be defined more by physical construction than by scientific discovery.
1
7
859
Morning headliners from both Axios and Tom Wigg today on the “AI CapEx Watch” as everyone is on their toes this week ahead of earnings.
We should have more info soon on Q2 AI CapEx levels, CEO commentary on ROI, and broader game theory dynamics playing out in the market.
1
1
8
2,057
I believe the answer is: You would take your time. You’d see how liquid cooling systems perform and alter your data center designs. You’d build new power assets in the right locations, and then build your data centers near fiber optic cables (2/11)
1
1
8
1,220
Sequoia Blog: sequoiacap.com/article/partn…
1
1
9
1,385
Today’s CapEx will likely yield fruit somewhere between late 2025 and early 2026, at which point we’ll find out if these larger models are intelligent enough to unlock new revenue streams and generate a return on investment.
1
8
823
Amazon Answer: Data Center Experience
Jassy: "Most companies deliver more capacity than they need...if you actually deliver too much capacity, the economics are pretty woeful...you can tell from...our operating income in AWS that we've learned over time to manage this."
8
2,682
3/ At that time, the hyperscalers were funding everything and taking on all the risk, driving the supply chain to greater and greater output. What’s changed this year is that Microsoft and Amazon have stepped back, leaving Oracle and Coreweave to fill the void.
1
2
8
1,642
WSJ Story: wsj.com/politics/national-se…
1
2
8
1,599
Meta Answer: AI Game Theory
Zuck: "There’s a meaningful chance that a lot of the companies are overbuilding now...[this is] a rational decision, because the downside of being behind is that you are out of position for the most important technology.”
bloomberg.com/news/videos/20…
1
8
1,649
2/ Last year, in “The AI Supply Chain Tug of War”, I wrote about pairwise game theoretic dynamics up and down the supply chain. “Each player wants to maximize profit while minimizing risk. This creates supply chain conflict, which lurks behind the scenes”
1
2
7
1,664
This changes AI in two fundamental ways:
1/ It changes the lead time between models. If before you could train your model in 6 to 12 months, now you need to add 18 to 24 months of construction time.
2/ It changes the source of maximum competitive advantage. In the new era, construction efficiency may matter more than research breakthroughs.
1
6
893
I meet with 200–300 new grads every year.
Their #1 question: Where should I work?
In the age of AI, the old algorithm (follow the smartest people a year ahead of you) is breaking.
Here's what I'm seeing 👇
2
8
897
The thesis is simple: Israel is one of the only Western countries with a draft. Its cadre of “technowarriors” have built some of the most elite startups in the world. As conflict becomes ever-more tech native, could we redeploy these technowarriors back into the defense sector?
1
1
8
716
In fact, the more you believe in AI, the more you might be concerned that AI model progress will outpace physical infrastructure. For example, once everyone has 100k clusters, big tech companies will need to figure out what to do with their 50k and 25k clusters (4/11)
1
6
1,091
Nvidia's Answer: CPU workloads moving to GPU
Jensen: "The answer is accelerated computing. We know that accelerated computing, of course, speeds up applications. It also enables you to do computing at a much larger scale."
1
1
7
2,161
When the history of this AI wave is written, the awesome scale of Microsoft, Amazon, and Google will be a big part of it. Without these companies and their fortress balance sheets – we would be significantly farther behind than we are today (9/11)
1
7
851
Full writeup on why we invested is here, with more details on the vision for fashionable eyewear as the ultimate interface:
sequoiacap.com/article/partn…
7
1,113
It’s been an absolute pleasure to partner with @AlonDror1 and @hamutalm. Alon is an elite Talpiot graduate, one of the top units in Israel. Hamutal previously ran Palantir’s Israel business. The Kela team that we’ve recruited over the last year is world class and growing.
1
7
527
3/ The deeper problem: Our economic growth has been fueled by cheap money and economic gains from globalization (see Edward Chancellor, Peter Thiel, Ray Dalio). If AI doesn’t deliver on its trillion-dollar promise, will another bubble emerge—or will this be the reckoning?
2
7
2,527
My rule has been: When I find an AI that’s actually enjoyable to talk to, we should invest. So many of these are painful and repetitive and lack memory. The conversation is choppy and it feels robotic.
1
6
456
1. AI will catalyze an energy transformation. New solar construction, battery innovation, a resurgence in nuclear energy—these will be long-term effects of the AI wave (2/7)
3
2
6
519
I've heard a few industry experts make comments along the lines of: No one will ever train a frontier model on the same data center twice—by the time the model has been trained, the GPUs will have become outdated, and frontier cluster sizes will have grown (5/11)
2
5
992
I knew within 10 minutes of using Sesame that it was different. It ended up getting over 1M users in a few weeks, and more than 5M conversational minutes. It was the first time an AI felt lifelike and dynamic, the tonality and cadence was incredible.
1
6
468
Kela is building a modern defense prime based in Israel, with a vision of defending not just Israel, but the entire Western Alliance.
Kela’s software platform and operational agility will enable Israel and its allies to better adapt in today’s fast-evolving defense landscape.
1
6
496
Excited to announce Sequoia's investment alongside @roelofbotha. You can find the private beta at Sesame.com/beta
1
6
1,332
2/ The theory: Each bubble has "bailed out" the economy from the last. When one asset crashes, another rises to keep the economy afloat. But with AI under scrutiny post-DeepSeek, the cracks are showing.
1
6
1,209
The cloud business today is a $250B market – the size of the entire SaaS sector, combined. The cloud giants are collectively worth more than $7T – this is one of the most powerful oligopolies in the history of business. Big tech is defending it's oligopoly (7/11)
1
1
6
836