Not many companies can lose $1trn in market cap. Nvidia (NASDAQ: NVDA) did, yet it’s still the second-largest company on the US stock market.

It was down 29% from January’s record high by market close on Tuesday (11 March). That’s worse than the Nasdaq, which has dropped 13% in three weeks.

Cheap chip maker?

Nvidia’s 1,665% gain over the past five years has been stunning. But even with that, the valuation still didn’t get close to the heights of some booming tech stocks of the past.

After this decline, forecasts have Nvidia stock on a prospective price-to-earnings (P/E) ratio of 25 for the 2025-26 financial year. For the year after, expected earnings rises would lower it to under 20.

On that score, Nvidia looks better value than the UK’s big growth champion, Rolls-Royce Holdings, with its forecast P/E of 31. And without meaning to downplay Rolls-Royce’s outlook, I feel the global demand for AI chips could easily outstrip aero engines in the coming decades.

Early mover disadvantage

Nvidia is seeing huge demand for its chips at a time of severe shortages in the power needed for today’s data centre surge. That means it can pretty much name its price and secure fat profit margins.

But when I think a stock price has skyrocketed partly due to an imbalance between supply and demand, I get a bit nervous. History tells us that markets are very good are leveling imbalances. The rest of the world’s chip makers are racing to catch up.

Are the big server developers puzzling over how best to use their capacity? Did they pile in because they can’t risk being left behind? Will the winners be the ones that successfully replace volume with efficiency? I think it’s a partial yes to all of these.

I don’t know who’ll be leading the AI chip business in 10 or 20 years. But I expect competition will be fiercer and margins could be a lot lower. Buying an early mover in the days after the first phase of growth can sometimes be a bad move.

How much growth

Even after Nvidia’s success so far, it still smashed through estimates in its fourth-quarter report last month. Revenue climbed 78% year on year, with earnings per share (EPS) up 82%.

At the time, CEO Jensen Huang described demand for Blackwell chips as amazing. He said “AI is advancing at light speed as agentic AI and physical AI set the stage for the next wave of AI to revolutionize the largest industries“.

But all this sky-high optimism reminds me of the old dot com bubble. Everyone thought the internet would advance at light speed, and they were right. But most really didn’t know how, and many of the early movers are forgotten names today.

If it gets it right, Nvidia could become the Amazon of the AI revolution. But even Amazon crashed 90% when the first wave collapsed, before powering back to become a long-term multibagger. Still, Nvidia’s modest P/E valuation makes me think any possible further losses probably won’t be anywhere near as bad as that.