Nvidia (NASDAQ:NVDA) is posting the kind of numbers that once justified a sky-high price tag. Yet the market now pays far less for each dollar of its expected profit than it did, on average, over the past five years. Something has to give. Either the earnings engine is about to stall, or the stock’s discount is out of date.
For anyone who hasn’t followed the story, Nvidia designs the chips and systems that train and run artificial intelligence models. Its graphics processors started out in gaming PCs, but today the money comes overwhelmingly from data centers run by cloud giants, AI labs, and governments. It increasingly sells entire racks of processors, networking gear, and software rather than individual chips. Which is it? For investors weighing companies featured among 10 Best Return on Equity (ROE) Stocks to Buy, the answer depends less on Nvidia’s record results than on whether its biggest customers can keep paying for them.
More Than a Fast Chip
Nvidia’s edge is the whole stack. Developers have spent years building on its CUDA software platform, and moving to a rival means rewriting code and re-tuning systems. That switching cost keeps buyers loyal even as competitors improve.
The numbers back it up. In fiscal Q2 2027 (ended in late July), Data Center revenue hit $89.0 billion, up 117% from a year earlier. GAAP gross margin reached 75.0%, compared with 72.4% in the year-ago quarter. A company losing its grip on customers usually has to cut prices; Nvidia’s margins went the other way, reflecting the formidable pricing power and balance sheet strength showcased among 10 Stocks With Low Debt and Strong Free Cash Flow.
Rubin Shows Up on Time
The newest proof is the Vera Rubin platform, which Nvidia says is now in full production. Rubin racks are already running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius.
Total revenue reached $96.2 billion in fiscal Q2, up 106% year over year and 18% from the prior quarter. Management guided fiscal Q3 revenue to $108.0 billion, plus or minus 2%. That outlook assumes no Data Center compute revenue from China, so any reopening there would be extra.
Nvidia is also pushing into new territory, with its Vera CPU built for AI agents and the Groq 3 LPX inference accelerator (a chip for running models rather than training them) in full production. The moat looks wider, not narrower.
Can Customers Keep Paying?
Here’s the catch. According to CNBC, Nvidia, Oracle, CoreWeave, and other AI stocks fell recently after a report on OpenAI’s revenue. That reaction gets at the core worry: a small group of buyers is spending enormous sums well ahead of the revenue AI might eventually generate.
Nvidia is even helping finance the buildout. It has set up compute financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, aiming to mobilize more than $500 billion in third-party capital, subject to definitive agreements. Bulls see that as clearing a bottleneck. Skeptics see a supplier propping up its own demand, a question that hangs even more heavily over specialized AI cloud providers. For now, navigating this deployment payoff remains the main risk to watch across the 11 Best Agentic AI Stocks to Buy Right Now.
Priced Like the Pack
At a forward P/E of 24.62, as of October 9, investors are paying roughly $25 for every $1 of expected earnings. That’s well below Nvidia’s 5-year average of 42.13 and only a whisker above the sector’s 23.35.
Now set that against growth. Adjusted EPS (which excludes items Nvidia treats as outside day-to-day operations) came in at $2.22 in fiscal Q2, up from $1.01 a year earlier. GAAP EPS was higher at $2.46, but the adjusted figure gives the cleaner read on the core business. Looking further out, expected EPS growth stands at 70.90% in 2028.
A company growing that quickly would normally command a premium to its sector, not near-parity. The market is effectively pricing in a sharp slowdown, which ties straight back to the customer-spending worry. That near-sector multiple also invites comparison with the foundry that manufactures its chips, highlighting why it remains closely watched alongside the other 10 Best AI Stocks to Buy Before They Explode.
Professional money hasn’t been scared off. Hedge fund interest ticked up, with 285 funds holding the stock in the most recent quarter, up from 275 in the prior one. Short interest sits at just 1.27% of the float, a sign of relatively limited bearish positioning.
The Discount Is the Debate
The evidence points to a business whose moat is widening while its valuation assumes the opposite. If Rubin keeps ramping and fiscal Q3 lands within its own outlook range, a multiple near the sector’s looks hard to defend for long.
That setup suits investors with a multiyear horizon who can stomach sharp swings tied to AI spending headlines. What would change the picture is clear: a pullback in spending by the big AI labs and cloud providers, or gross margin slipping meaningfully below the roughly 74% Nvidia has guided for.
READ NEXT: This 6.3% Dividend Yield Looks Too Good to Ignore. Is the Stock a Buy? and 10 Best 52-Week Low Technology Stocks to Buy According to Analysts.
