The interesting thing about Nvidia Corporation is not that the business is doing well. That is the boring thing. The interesting thing is that almost every other publicly-traded technology business on the Tape — Microsoft, Alphabet, Meta, Oracle, Amazon — is doing well, in some meaningful part, becauseNvidia is selling them the hardware to do well with. Nvidia's revenue is, in 2026, a leading indicator of capital spending across the entire artificial-intelligence build-out. Mr. Market understands this perfectly. The stock has compounded at well over fifty percent a year for half a decade.

The protocol, on the other hand, finds Nvidia a difficult business to reckon with. The forward analyst consensus on earnings growth is enormous — well above the twenty-percent cap the protocol allows. The historical growth rate over the trailing five years is even larger. The maintenance-CapEx proxy is unstable because Nvidia is in the middle of a capacity expansion the like of which the semiconductor industry has not seen since the late 1990s. Owner Earnings, in any honest sense, are growing at thirty or forty percent per year, and the protocol, in any honest sense, cannot model that without producing valuations that look more like science fiction than valuation.

The protocol, on the other hand, finds Nvidia a difficult business to reckon with.

So we apply the rule we apply to every business of this kind. The technology sector premium adds two percentage points to the discount rate. The analyst-consensus growth is capped at twenty percent. The fade schedule pulls the long-run growth rate gently down to two-and-a-half percent over a decade. The resulting intrinsic value is what it is — meaningfully above today's asking price, but not by the eye-watering margins one might naively expect from a business growing at fifty percent. The discipline of the protocol is, in part, an antidote to the discipline of momentum.

We do not, frankly, know what Nvidia is worth. Neither does anyone else. The honest answer for a business compounding at fifty percent a year is that any number we produce is a guess across a range that spans an order of magnitude. The protocol's guess is somewhere in the middle of that range. Mr. Market's guess, this morning, is somewhere in the upper half. Reasonable people, including reasonable people on this masthead, disagree on whether Mr. Market is right.

What we will commit to is this. If Nvidia is, in 2030, still the single business that every other technology business depends on, the implied compounding will have paid for itself many times over. If Nvidia is, in 2030, one of three or four businesses competing for that same market share — which is what semiconductor markets historically do — the compounding will not have paid for itself, and the buyer at today's price will look back at this dispatch and conclude they should have read it more carefully. We do not know which world we are in. Neither does Mr. Market. He is, however, asking us to commit to one of them.

For the patient owner, the call on Nvidia is, more than for any other business on the Tape, a call about what one believes about the next ten years of artificial intelligence economics. The protocol does not have a strong view on that. The editors do not have a strong view on that. We have, instead, a number, applied consistently, with the caveats clearly disclosed. That is the publication.

— The editors