It’s no secret Nvidia (NASDAQ:NVDA) stock has absolutely crushed the S&P 500 over the past decade. But the level of outperformance is truly jaw-dropping, as the artificial intelligence (AI) chip supremo has left others in a trail of dust.
| 10-year return (excluding dividends)* | |
| Nvidia | 22,874% |
| Advanced Micro Devices (AMD) | 7,763% |
| Comfort Systems USA | 4,812% |
| Micron Technology | 4,025% |
| Arista Networks | 3,633% |
All these businesses have had rocket boosters put under them from the AI infrastructure buildout. Not just chips, but also the high-speed networking switches (Arista Networks) and cooling systems (Comfort Systems) needed for data centres.
But to put Nvidia’s return into context, a £7,500 investment made a decade ago would now be worth over £1.6m. And while the same investment made 18 months ago wouldn’t be anywhere near as dramatic, it would still have turned £7,500 into around £11,500, even after a slight strengthening of the pound.
More recently however, the stock’s performance hasn’t been strong. In fact, Nvidia is at the same level as August 2025. So might there be an opportunity for investors to look at here?
From chatbots to agents and robots
Since ChatGPT was released in late 2022, Nvidia’s data centre revenue has risen 13 times, from $15bn in fiscal 2023 to almost $194bn in fiscal 2026. And in this time, the firm’s net profit margin has ballooned from 16.2% to a barely believable 55.6%.
Jensen Huang, Nvidia’s leather jacket-clad founder and CEO, said in February 2023: “AI is at an inflection point…From startups to major enterprises, we are seeing accelerated interest in the versatility and capabilities of generative AI“.
In February, he wrote: “Computing demand is growing exponentially — the agentic AI inflection point has arrived“. Now, what he’s talking about here is autonomous AI agents that can reason and take action. And Huang says business adoption of these is “skyrocketing“.
Running these agents requires a massive amount of inference (ie using a trained AI model to make predictions, decisions, or generate content on new data). And Nvidia’s latest Rubin platform delivers up to a 10 times reduction in inference token cost compared with the existing Blackwell platform.
So in the space of three years, we’ve gone from one inflection point (generative AI) to another (agentic AI). But management sees yet another on the horizon — ‘physical AI’. This includes humanoid robots and self-driving vehicles.
Last year, Nvidia’s nascent Automotive and Robotics division saw revenue jump 39% to a record $2.3bn. And CFO Colette Kress recently said: “Robotaxi rides are growing exponentially, with commercial fleets from Waymo, Tesla, Uber, WeRide, and Zoox, and many others expected to scale from thousands of vehicles in 2025 to millions over the next decade, creating a market poised to generate hundreds of billions of dollars of revenue.”
The key takeaway here is that the AI age is just getting started.
Lucrative opportunity?
But that doesn’t mean there aren’t risks. A big one is high customer concentration, with the top five cloud providers (the hyperscalers) accounting for just over 50% of Nvidia’s data centre revenue. If these suddenly cut back on AI spending, the firm’s growth would slow.
Yet the stock’s trading at just 17.5 times next year’s forecast earnings. This is a discount to the Nasdaq100 index and its own 10-year average, suggesting Nvidia’s currently undervalued.
At $183, I think the stock’s worth a closer look.
