When the CEO extends the company’s $500 billion revenue projection for 2026 by adding another $500 billion for 2027–we’re talking $1 trillion in revenue here, folks–and the stock barely budges, I think we’re looking at a stock in need of a valuation reset.
That’s exactly what happened to shares of Nvidia (NVDA) yesterday and today, March 16 and 17.And it’s why I’m selling Nvida out of my 50 Stocks Portfolio tomorrow, March 18. I expect I’ll be back into Nvidia shares when the valuation is less stretched–either because of a pull back in the shares or because projected revenues have turned into booked earnings (or some of each.) I’ve done this rotation in and out of Nvidia once before, selling in late 2023 and then rebuying in December 2023 for a 290% gain as of the close on March 17, 2026. Nvidia shares were dawn 1.75% for 2026 as of the close on March 16.
Here’s what has happened to Nvidia’s share price in the last two days. On March 16, Nvidia closed up 1.65%. On March
17, Nvidia closed down 0.7% on the day.
For context, on Monday, March 16, the Nasdaq Composite gained 1.2%. On March 17, 2026, the index gained 0.6%
So on March 16 Nvidia slightly outpaced its index. On March 17, the shares trailed the Nadaq Composite.
.
And here’s what happened at the company.
CEO Jensen Huang, addressing crowds at the company’s big annual GTC event, unveiled a variety of new products while predicting that its flagship AI processors would help generate $1 trillion in sales through 2027.
During a 2 1/2-hour keynote address, Huang announced plans to push deeper into central processing units–Intel’s home turf–and introduced chips made with technology acquired from startup Groq. In a typically restrained presentation Huang also said Nvidia was developing chips for data centers in outer space. “I believe that computing demand has increased by 1 million times in the last two years,” he said. “It is the feeling that we all have. It is the feeling every startup has.”
It’s not that I don’t believe Huang. Nvidia indeed has a product roadmap to back up his enthusiasm.
Nvidia guided that purchase orders tied to Blackwell chips and Vera Rubin chips could reach roughly $1 trillion through 2027, effectively doubling a prior $500 billion estimate.
It’s just that the market’s immediate response to these projections is saying that when your stock trades with a valuation of $4.4 trillion, a $1 trillion in sales by the end of 2027 isn’t enough.
Especially when it’s pretty easy to construct a “risk narrative” for the AI sector that argues some of this revenue won’t actually materialize. It’s not outlandish to imagine that some of the AI and data center companies that have booked orders with Nvidia–because if you don’’t book now, you risk not being able to get any Nvidia chips at all since supplies are so tight–might not be actually take delivery.
There are also risk factors such as the fact that AI companies are not profitable right now. That their incredibly ambitious and expensive capital spending plans depend on a continued ability to raise billions in the debt market. That some recent chip purchase deals seem to be very circular with a chip maker such as Nvidia essentially financing the purchase of its own chips, often in exchange for equity.
And, of course, Nvidia’s chips do face increasing competition from U.S. chip makers such as Advanced Micro Devices (AMD), from data center/cloud companies such as Alphabet/Google (GOOG) and Amazon (AMZN). And from government-funded Chinese chipmakers.
I’m not counting Nvidia out by any means. The company has responded to increased competition by accelerating its technology development schedule to replace its entire product lineup on an annual basis while adding new components. The next design of its flagship AI processors—-Vera Rubin–is set to appear in the second half of 2026. (Vera Rubin was a pioneering astronomer whose observations provided evidence for the existence of dark matter.) Rubin will be followed by a generation named after Richard Feynman, the American physicist who died in 1988, with customized high-bandwidth memory.
Still, I’d like get this Nvidia story at a price that reflects some of the risks facing the company.
