David Cahn

Cahn’s “steel, servers and power” thesis landed: AI’s operative unit moved from dollars toward gigawatts, making power his “best trade of 2025” and construction execution a moat. Generators were sold out until 2030, electricians were being flown to Texas, and AI construction became a material contributor to US GDP. Yet the end demand remains unresolved: his original $600 billion revenue question has become roughly $840 billion.

AI can be a civilization-scale technology and still be a bubble whose compressed timeline “will incinerate capital.” Cahn expects AI to transform society over 50 years, but markets are financing that outcome as if it must arrive quickly on today’s specific chips. The investor’s task is therefore survivorship: find companies with customer love that can endure volatility, not businesses dependent on infinite cheap capital.

The cleanest bubble winners are consumers of compute, because excess capacity lowers their COGS and raises gross margins; producers inherit commodity economics. Harry challenged that framing with AWS, Azure and Google Cloud, but Cahn argued those monopolies were built before their opportunity was obvious. AI’s value is visible to everyone, inviting competition and making monopoly profits less likely—good for consumers, difficult for capacity owners.

The system’s clearest fragility is the transfer of risk from Microsoft and Amazon to smaller operators and then back to chip suppliers through circular financing. Oracle and CoreWeave cannot absorb hyperscaler-scale risk, while chip companies can fund projects cheaply because the spending returns as booked revenue. With one gigawatt costing about $40 billion—or $50-$60 billion on Vera Rubin—the announced 100-250 gigawatt ambitions become unfunded multi-trillion-dollar questions.

The most important overestimate is timing, not AI’s eventual importance. Andrej Karpathy’s “decade of agents,” Richard Sutton’s doubts about the current paradigm, Ilya Sutskever’s “pre-training is dead,” and Sam Altman’s “gentle singularity” all cut against status-driven lab chatter about AGI in 100-300 days. If the breakthrough arrives on Feynman-era chips in 2028 or a decade later, today’s H100 and B100 warehouses still bear the loss.

Neither premium venture brands nor abundant capital can manufacture product-market fit: “capital is fuel, but capital does not create the engine.” Cahn accepts that Sequoia can improve recruiting and marginally change probabilities, but “you can’t make a company succeed”; Profound was already ripping before Sequoia invested. The same discipline applies to metrics: margins can improve from 30% to 70%, while today’s strongest adoption signal is the “zero to 100 club”—not a rigid threshold, but evidence of unusually strong demand.

Defense may be “the next AI,” but it will produce a few national champions rather than a broad SaaS-like ecosystem. Harry challenged Sequoia’s absence from Helsing and Anduril; Cahn conceded Sequoia was late to defense, said the sector is roughly two years after the Transformer paper and before its ChatGPT moment, and estimated it is only about “1%” through a 50-year catch-up. He sees Anduril in the US, Kela from Israel and Stark in Europe as candidates in a market where deterrence—not celebrating “cost per kill”—is the objective.

🔗 Original source & video: Sequoia Partner, David Cahn on Who Wins in AI, Defence & The New $0–$100M Playbook

David Cahn reads DeepSeek as evidence that China reached GPT-4 parity, not as an unexpected frontier breakthrough. The more consequential signal, he argues, was Ilya Sutskever’s suggestion that “pre-training is dead” or “scaling laws are dead”: competitors were always going to copy the existing generation, while the source of the next leap remains unknown. His ranking is explicit: “DeepSeek maybe is overhyped, and what Ilya said is maybe a bit underhyped.”

Cheaper models are bullish for AI applications even if they threaten scarcity-based infrastructure assumptions. Lukas Biewald notes that decades of falling compute costs have usually driven greater aggregate usage; Cahn likewise calls cheaper inference “great for the application layer” and says the market “sort of freaks out at the wrong moments,” reacting to commoditization that was visible six months earlier. Satya Nadella’s longstanding argument—that models would be distilled and become much cheaper, benefiting hosts such as Microsoft—now looks central.

Stargate’s $500 billion ambition and DeepSeek’s efficiency represent competing capital-allocation regimes. Stargate assumes larger data centers keep producing better models, while DeepSeek points toward smaller models and value migrating upward to applications. The revealing change for Cahn is Microsoft stopping at its existing $80 billion data-center commitment, which may mark a shift from hyperscaler balance-sheet funding to “credit-funded data centers or leveraged data centers.”

Cahn’s $600 billion question remains unresolved because one year of infrastructure investment requires an enormous downstream revenue base. His napkin math starts with roughly $150 billion of NVIDIA GPU run-rate revenue, doubles it to $300 billion for power and data-center infrastructure, then doubles it again so applications can earn 50% gross margins. Another equivalent investment year takes the implied obligation to $1.2 trillion—“almost this debt that we’ve sort of invested in that we now have to go pay back over time.”

Hyperscaler capex is stabilizing, but AI revenue has not caught up. Cahn estimates Microsoft at roughly $20 billion per quarter and Google at $13 billion, with Amazon likely stabilizing in the low $20 billions and Meta in the low teens; that keeps the question from immediately becoming a trillion-dollar question. Yet OpenAI still represents the lion’s share of ecosystem revenue, leaving something close to the earlier “$500 billion hole.”

The spending persists because the cloud oligopoly is trapped in a strategically rational prisoner’s dilemma. A roughly $500 billion cloud market is the “golden goose” for Microsoft, Amazon, and Google, while seven companies account for 33% of the S&P 500; existing profits therefore finance the race. Executives believe the opportunity is immense, but their defensive logic is equally important: “We have to spend, or we’re going to fall behind.”

Cahn’s clearest application-layer thesis is profession-specific AI search built around how users actually think. He uses Perplexity 10–20 times daily and sees parallel products for lawyers and doctors, with differentiation across intent extraction, proprietary data, answer formatting, and “cognitive architecture.” The opportunity is an “AI search pal” mapped to a profession’s problem-solving patterns, not merely a generic model with a new interface.

His venture framework favors founders and problems he can still believe in when fashion reverses, while demanding far more than technical superiority. A product that is “20% better” may be commercially irrelevant if a busy buyer cannot see why switching matters; conversely, patience paid off at Weights & Biases, and deep conviction enabled early bets on Runway and Form Energy. That same search for substance culminates in AI’s deepest uncertainty: whether it becomes merely superhumanly intelligent or also self-reflective and conscious—“fundamentally, that’s a religious question in many ways.”

🔗 Original source & video: Deepseek, Stargate and AI’s $600 billion question with Sequoia Capital’s David Cahn