With a market cap of about $4.5 trillion, Nvidia (NVDA 1.32%) is the largest company on the planet. However, it might also just be the most undervalued artificial intelligence (AI) stock right now, too.
But don't a lot of pundits say Nvidia is overvalued? That's true; however, most claims that Nvidia is overvalued stem from its trailing price-to-earnings (P/E) of around 45.5 times, which on the surface is high. Yet, based on 2026 analyst estimates, its forward P/E is below 25 times, and its price/earnings-to-growth (PEG) ratio is below 0.7 times (with below 1 times considered undervalued).
But that's not all. The company also carries around $52 billion in net cash and securities on its balance sheet, and it's on pace to generate around $85 billion in free cash flow this year. For a company growing as quickly as Nvidia, those valuation metrics are cheap.
NASDAQ: NVDA
Key Data Points
The AI infrastructure leader
Why Nvidia is perhaps the cheapest AI stock in the market is also directly intertwined with its growth. The company has been a growth machine. Last quarter, it grew its revenue by 62% year over year, and its revenue was up nearly tenfold from just two years ago. Its adjusted earnings per share, meanwhile, climbed 60% year over year.
Nvidia also doesn't see growth slowing down in the near term. For fiscal Q4, it forecast that its revenue would climb 65% year over year to $65 billion. Its outlook for 2026 also looks promising, as the three major cloud computing companies have all indicated that they will spend aggressively on data infrastructure next year, as have others, including Meta Platforms and OpenAI.
Meanwhile, Nvidia could see its sales get a further lift in 2026 after the U.S. agreed to let the company sell its H200 chips to approved commercial customers in China. The Trump administration earlier banned the export of Nvidia's H20 chips to China, but in a turnaround, it will now allow the even more powerful H200 chips to be sold, in exchange for the U.S. government getting a 25% cut.
Over the medium term, Nvidia has projected that data center capital expenditure (capex) could hit $4 trillion by the end of the decade. As the main provider of the chips that power AI workloads, it is in a prime position to capture more than its fair share of this spending, as a large percentage will go toward chips and networking components.
Image source: Getty Images.
Nvidia's edge comes from the ecosystem it has built around its chips, which are called graphics processing units (GPUs). The name stems from the chip's original purpose, which was to speed up graphics rendering in video games. However, to expand the use cases for its chips, Nvidia developed the CUDA software platform to let developers easily program its chips for other tasks.
This was a smart move, but the even cleverer one was giving it away for free and seeding it into top universities and research labs that were doing very early work on AI. As a result, nearly all foundational AI code is written on CUDA and optimized for its GPUs. The company didn't stop there, getting into networking and creating the proprietary NVLink interconnect systems that let its chips quickly share data and memory, essentially helping them act as one big unit.
As a result, Nvidia now holds over 90% market share in the data center GPU space, which is one of the largest and fastest-growing markets ever seen. While it is seeing some increased competition, its scale, ecosystem, and the flexibility of its chips are unmatched, positioning it to continue to lead the market higher in the decade to come.
I could see Nvidia potentially earning around $20 per share in fiscal 2030 (ending January 2023) if revenue growth just continues to step down gradually.
| Metric | FY2026E | FY2027E | FY2028E | FY2029E | FY2030E |
|---|---|---|---|---|---|
| Revenue | $213 | $320 | $463 | $649 | $876 |
| Gross profit | $155 | $233 | $338 | $473 | $639 |
| Adjusted operating expense | $21 | $27 | $35 | $46 | $61 |
| Operating Income | $134 | $206 | $303 | $427 | $578 |
| Net income | $114 | $175 | $258 | $363 | $491 |
| Adjusted EPS | $4.70 | $7.21 | $10.61 | $14.95 | $20.22 |
Data source: Author's projections. Numbers are in billions.
If that's the case, it's one of the cheapest AI names out there and has a lot of upsides ahead, including in 2026. As long as the AI infrastructure boom continues, it is a stock to own.
