Anthropic's $2T IPO Math Doesn't Add Up: Patrick Boyle Does the Maths

Patrick Boyle, one of YouTube's sharpest finance educators, just did something radical: he did the actual math on Anthropic's reported $2 trillion IPO valuation target. The result is not flattering. Even under the most absurdly optimistic assumptions, including zero operating costs, zero taxes, and zero employees, every dollar of Anthropic's revenue discounted at the current Treasury rate is worth roughly $1.27 trillion today. That's a $730 billion gap between the ask and any defensible number. That's not a rounding error. That's a different zip code.

Here's where it gets interesting, because the AI story is not uniformly broken. Boyle singles out $NVDA as the one company in the AI ecosystem that actually passes a basic valuation sanity check. Nvidia trades at roughly 17x forward earnings, its cheapest valuation in over a decade, despite growing sales from $27 billion to an estimated $410 billion over four years. That is not bubble math. That is a business growing into its price tag at warp speed.

Meanwhile, Seeking Alpha's PEG ratio breakdown adds useful texture here. Nvidia's PEG ratio sits at approximately 0.23, using roughly 29x earnings against ~125% growth, making it one of the few mega-cap tech names that looks genuinely cheap on a growth-adjusted basis. Apple, for comparison, clocks in at a PEG of 1.19 (38.7x P/E, 32.6% growth), which reads as slightly overvalued on the same framework.

The broader tension the Boyle video captures is this: the AI funding ecosystem has decoupled from any tether to conventional valuation. When a private company can target a $2 trillion IPO valuation that exceeds what its entire future discounted revenue stream can justify, the market is pricing in something that cannot be quantified. Faith, FOMO, and a lot of venture capital relationships are doing a lot of heavy lifting right now.

Substrates, the unglamorous connective tissue of AI chips, are getting fresh attention on CNBC as the new gold of the semiconductor supply chain, as they become the actual bottleneck in AI compute performance. The real money in AI infrastructure may be flowing through components most retail investors have never Googled.

Anthropic may build the most important AI in human history. It may also be the most expensive story stock since WeWork convinced people a real estate company was a tech company. The difference is, this time the numbers are bigger and the hype is louder.