On June 20, 2024 David Cahn, a partner at the venture firm Sequoia Capital, published an essay titled “AI’s $600B Question.” It was a follow-up to his September 2023 piece “AI’s $200B Question,” which had asked a deceptively simple thing about the generative-AI boom: “Where is all the revenue?” In the year between the two essays, the gap he was measuring had grown, in his framing, from a 200 billion dollar question to a 600 billion dollar one.
Cahn’s method was a back-of-envelope chain tied to NVIDIA’s data-center sales. He took NVIDIA’s run-rate revenue forecast, multiplied it by two on the reasoning that GPUs are only about half the total cost of an AI data center (the rest being energy, buildings, and supporting hardware), then multiplied by two again to reflect a roughly 50 percent gross margin that the end users of that compute would need to earn. The result was an estimate of how much annual revenue AI products would have to generate to justify the infrastructure being built - and that number, set against actual AI revenue, left a very large hole.
Cahn was careful not to call AI a fraud or a fad. He wrote that AI “is likely to be the next transformative technology wave” and argued that builders and founders would ultimately benefit from cheaper compute and accumulated learning even if the current investment cycle overshot. His warning was aimed at speculative excess and at investors who assumed the revenue would simply appear. The essay became one of the most-cited reference points in the subsequent debate about whether AI spending had run ahead of demand.
Why a business reader should care: this piece turned a vague unease about AI hype into a specific, repeatable calculation, and it set the terms for the bubble debate that followed - the question of whether hundreds of billions in compute spending will be matched by the revenue needed to pay for it.