Nvidia’s Stock Price Three Years From Now: A Prediction That Is Making Wall Street Argue

In keynote speeches, Jensen Huang frequently makes statements that, at the time, sound like polished corporate optimism but later prove to be accurate. This was also the case at the GTC 2026 conference in March. Huang informed analysts and investors on stage in San Jose that by the end of 2027, the combined sales of Nvidia’s Blackwell and Rubin chip architectures would total $1 trillion. The number was absorbed by the room. In the same year, Wall Street had been projecting $480 billion. This disparity, which is approximately $520 billion, is at the core of a debate that is currently taking place on trading desks, in research reports, and during earnings calls.

Over the previous three years, Nvidia’s stock has increased by about 525%. That’s not a typo, and it’s not just speculative fever. The business has regularly produced revenue figures that were higher than analysts had predicted, sometimes by sizable margins. Huang predicted $500 billion in total sales of AI chips for 2025 and 2026 at the GTC conference last year. At the time, analysts were using more conservative projections. It’s a well-known pattern: Huang gives a figure, Wall Street models a lower figure, and Huang’s figure ultimately approaches reality. Even though the difference between his trillion-dollar prediction and the consensus is nearly comical, some investors take it seriously because of this history.

For a company the size of Nvidia, the current valuation calculation is out of the ordinary. With a forward price-to-earnings ratio of about 15.7x and a share price of about $177, the stock is trading at a multiple that, in some ways, appears less expensive than the S&P 500. That may seem odd for a $4.3 trillion company, but it shows how quickly earnings have increased in comparison to the movement of the stock price. After two consecutive quarterly losses, shares are down from their peak of $212. This is a difficult period by any measure, but when compared to the longer trajectory, it appears more like a pause than a breakdown. It’s difficult to ignore the fact that every time Nvidia seemed to stall, the question of whether the cycle of AI spending was coming to an end turned out to be premature.

The bull case, which was developed by analysts at Motley Fool and other companies using comparable models, goes something like this: if Nvidia can maintain yearly revenue of about $600 billion by fiscal 2029, if it maintains a profit margin of about 50%, and if the market assigns a relatively modest 30 times trailing earnings, the stock will land somewhere around $370. From the current price, that would be more than a doubling. Although there is potential for compression as competition grows, Nvidia has maintained pricing power in ways that most semiconductor companies are unable to, so the margin assumption isn’t irrational. The company’s gross margin is currently above 71%. AMD is making an effort. Microsoft, Google, and Amazon are expanding their custom chip programs. However, none of them currently pose a threat to Nvidia’s position in the manner that some pessimists anticipate.

Although it is quieter, the bear case is still important. Some analysts believe that the buildout of AI infrastructure is front-loaded, that hyperscalers like Google and Microsoft are currently spending at rates that reflect multi-year demand compressed into a short window, and that as data centers constructed in 2024 and 2025 come online, the rate of Nvidia chip purchases will normalize. Huang responds to this by saying that construction costs, such as those of the actual buildings, power systems, and cooling infrastructure, currently account for a sizable amount of capital expenditures rather than chips. Spending should return to hardware and, consequently, to Nvidia once construction is finished. It’s a plausible argument, and it might happen just as he says.

In its coverage of the GTC event, Barron’s effectively conveyed the tension, pointing out that 93% of the roughly 70 analysts covering the stock have buy ratings, with an average price target of about $267, suggesting a roughly 45% increase from current levels. Depending on how you interpret it, such near-unanimous bullishness can either be an indication of sincere conviction or a red flag. Unanimity is rarely rewarded by markets. The question of who still needs to be converted and what happens when the next disappointing data point shows up arises when everyone agrees that a stock is a buy.

Observing Nvidia’s stock over the previous two years gives the impression that the company has progressed beyond the stage where the story can be determined by a single quarter’s performance. A type of embedded demand that is challenging to swiftly reverse is created by the infrastructure being built around AI, such as the data centers, power capacity, and cooling systems going into facilities outside of Phoenix and in the Virginia suburbs. Orders for chips must be canceled along with the facilities they were intended to fill. As a result, Nvidia has an advantage over the majority of fast-growing tech firms: a company with physical obligations that span years rather than quarters.

Whether the stock hits $370, something closer to $267, or something completely different is likely to depend more on whether the AI spending cycle lasts as long and as deep as Jensen Huang has insisted it will for the past two years than on any one analyst’s model. The smart money is on not betting against him, according to history, at least recent history.