Sequoia partner: AI must earn $3 trillion to justify spend

Sequoia partner David Cahn calculates that the AI industry will have to earn $3 trillion to justify all those chips and other data center expenditures — the flip side of a projected $1.5 trillion in AI infrastructure spending for 2026. That required-revenue figure is probably an underestimate, he notes, since rising memory costs and increasing use of exotic or inference-specific chips will likely push it higher still.

The estimate marks an escalation from three years ago, when Cahn first did the math on Silicon Valley’s AI infrastructure spend. In 2023, reacting to Nvidia’s reported annual GPU revenue of $50 billion, he calculated that $200 billion in revenue would be required to pay back the up-front investment. Cahn, the Sequoia partner, writes that “Recently, the required revenue per GW of CapEx has sharply increased due to these bottleneck dynamics and rising costs of construction.”

On the revenue side of the ledger, the gap is still wide. Anthropic is thought to have hit $60 billion in annual recurring revenue (ARR — a standard measure of subscription revenue run-rate), while OpenAI reportedly earned $13 billion in 2025; the company said in November 2025 that it had reached $20 billion in ARR.

Apollo (the asset manager) chief economist Torsten Slok is watching that gap closely. He points out that the hyperscalers — Google, Meta, Microsoft, and Amazon, the cloud providers whose spending underpins the AI buildout — are all predicting massive accelerations in their free-cash flow in 2028. Slok also flags a countervailing trend: more organizations are turning to cheaper open weight models, often Chinese, rather than those from the frontier labs, while overall token prices fall. OpenAI’s latest model, according to CEO Sam Altman, is 54% more token efficient on coding tasks — good news for users, but a potential squeeze on companies whose revenue depends on rising token usage if that usage doesn’t grow to compensate.

If hyperscalers miss their 2028 cash-flow targets, Slok warns, the fallout could extend well beyond the tech sector: “with so much riding on so few names, a slower payoff wouldn’t just be a sector problem, it would risk tipping the economy into recession and the S&P 500 (the US stock market index) into a correction.”

Why it matters

The gap between $1.5 trillion in AI infrastructure spending and the actual revenue generated by top AI startups is a macroeconomic risk, not just an industry one — if hyperscalers don’t see the payback they’re forecasting by 2028, the fallout could extend to the broader economy and stock market.