Nvidia’s Next Act – Predicting the Trillion-Dollar Giant’s Stock Price 3 Years From Now

This year, the San Jose convention center had a different vibe. It can only be described in that way. There was a certain energy that existed between a product launch and a revival meeting as I entered Nvidia’s GTC 2026 conference, past the massive screens that cycled through renderings of data centers and robotic systems.

Standing on stage in his now-iconic black leather jacket, Jensen Huang said something that would have seemed unreal to most people: by the end of 2027, the combined revenue from just two chip families would reach $1 trillion. The audience reacted as they usually do at these kinds of events.

However, Nvidia’s stock hardly moved in the real market. The most intriguing aspect of Nvidia at the moment is probably that divide between the spectacle inside and the apathy outside.

When the numbers are taken at face value, it is truly difficult to dispute them. Over the last three years, Nvidia’s stock has increased by about 525%, making it one of the best performers of the current technological era. Almost entirely due to the demand for AI computing infrastructure, revenue increased from about $27 billion in fiscal 2023 to over $120 billion more recently.

For the time period Huang mentioned, Wall Street analysts had projected total data center revenue of about $965 billion; his forecast easily exceeded that ceiling. Notably, there is a pattern here. Nvidia’s growth rate has been consistently underestimated by the analyst community, year after year and quarter after quarter, to the point where it has practically become a running narrative in the financial media. Some investors believe that projecting Nvidia is now more akin to estimating the rate at which a wildfire spreads than it is to financial analysis.

And yet. Since January, the stock has essentially remained unchanged, with a price-to-earnings multiple that has fallen below the overall S&P 500, something Nvidia hasn’t experienced in more than ten years. It’s odd to see a business posting these kinds of figures.

The market cap, which is currently at about $4.4 trillion, creates its own gravitational problem: doubling at that size necessitates adding an additional $4.4 trillion in value, which starts to put pressure on even the most optimistic models. Many semiconductor investors are specifically searching for stocks where they can at least create a plausible scenario involving a double, according to a recent observation by Josh Buchalter of TD Cowen. Even if the underlying business continues to do well, it is becoming more difficult to sketch that scenario with Nvidia.

However, valuation math is not the more important issue. It concerns what the market for AI computing truly requires going forward. GPU chips that were incredibly well-suited for training large language models—basically, the first generation of significant AI infrastructure—were the foundation of Nvidia’s dominance. Major hyperscalers have been investing all of their free cash flow in data centers for the past two years.

At GTC, Daniel Newman of Futurum Research asked directly whether Nvidia has a real second act after this infrastructure cycle or if the $1 trillion estimate is more of a ceiling than a floor. A different hardware mix that relies more on CPU chips is required as AI models move toward inferencing, the stage where they actually respond to queries in real time. Although AMD and Intel, two dynamic companies, have long held this market, Nvidia now designs its own CPUs.

It’s difficult to watch this without thinking of Intel in 2012—dominant, extremely profitable, perched atop a stronghold of market share while the ground beneath it quietly changed. That comparison is probably unfair because Nvidia has proven to be more adaptable than Intel, but the parallel still exists. Huang’s announcement of the new Groq inferencing chip at GTC is precisely the kind of product that indicates Nvidia anticipates the shift and is working to be the first. The next three years will primarily focus on whether it truly makes it there and whether it can maintain its margins in the process.

If the company generates between $590 and $600 billion in revenue by fiscal year 2029 while maintaining a 50% profit margin, the stock price calculation at a reasonable multiple yields a figure that is significantly higher than where shares currently trade. According to some analysts, the market capitalization will be close to $8 trillion by 2028. Those forecasts might be accurate.

Additionally, it’s possible that competition from custom silicon manufactured by Google, Microsoft, and Amazon, along with a natural cooling of data center spending after the current construction wave ends, will put significant pressure on margins that the models haven’t yet fully priced in. Nvidia has outperformed every prediction made thus far. That record is authentic. However, it remains uncertain if the upcoming three years will yield the same degree of assurance that the previous three merited.

The thing that most amazes you when you watch all of this happen from a distance is how composed Jensen Huang seems when making such big claims. A trillion dollars in sales of chips. Declared at a conference in San Jose, briefly reported in the financial press, and then largely ignored by a market that has merely redefined what is newsworthy when Nvidia is speaking. The biggest risk of all may be that normalization—not that the business won’t deliver, but that delivering won’t be sufficient.