Nvidia Just Blew the Doors Off Wall Street — and the Entire AI Trade Now Hangs in the Balance

Nvidia delivered yet another blow out quarter Wednesday evening, a performance that not only crushed already-lofty expectations but also reasserted the company’s dominance at the dead center of the global AI investment cycle. Heading into the print, Wall Street was expecting roughly $1.25 in EPS on $54.8–55 billion in revenue—already a 50%+ year-on-year jump. Nvidia itself had guided to $54 billion (±2%) last quarter. A few aggressive analysts penciled in upside toward $56–57 billion. Nvidia skipped all of that and cleared the bat by a mile, posting $57.0 billion in revenue, up 62% year-on-year and 22% sequentially, and $1.30 in EPS, up 67% year-on-year and 20% sequentially. That marks not only another record quarter but one that decisively reaffirms AI infrastructure as the strongest spending theme in global markets.

The results arrive at a crucial technical moment. Nvidia shares had pulled back from an all-time high near $212 on October 29 and were sitting directly on the 50-day moving average—a level traders viewed as either a bounce zone or a trapdoor. With the stock now breaking cleanly above that moving average after the report, the question is simple: can Nvidia hold these gains? A post-earnings fade after numbers like this would be a psychological blow to the entire AI trade. But on fundamentals alone, Nvidia just delivered everything the bulls needed.

The data center segment—Nvidia’s economic engine and the focal point for every investor—was again the standout. Data center revenue hit a staggering $51.2 billion, up 66% year-on-year and 25% sequentially, fueled by what management called three platform shifts: accelerated computing, large-scale AI models, and the rapid emergence of agentic AI applications. This is where the Blackwell architecture cemented its leadership. Blackwell Ultra is now the company’s top-performing platform across every customer category, while the prior-generation Blackwell systems continue to experience brisk demand. If investors wanted hard proof that the AI build-out remains “insatiable,” they just got it.

Compute revenue inside the data center reached $43.0 billion, up 56% year-on-year and 27% sequentially, helped by accelerating deployments of GB200 and GB300 systems. But the real jaw-dropper once again was networking. Revenue in the networking segment surged to $8.2 billion, up 162% year-on-year and 13% sequentially, driven by explosive demand for NVLink, XDR InfiniBand, and the expanding Spectrum-X Ethernet stack used to interconnect GB200 and GB300 clusters. As supply availability improves, shipment cadence varied from last quarter, but the growth trajectory remains unmistakably vertical.

On the graphics side of the business, results were solid though less spectacular. Gaming revenue came in at $4.265 billion, up 30% year-on-year, but down 1% sequentially as channel inventories normalize heading into the holiday season. Professional Visualization revenue rose to $760 million, up 56% year-on-year and 26% sequentially, boosted by the launch of DGX Spark and rising workstation demand for Blackwell-powered systems. Automotive continued its steady climb to $592 million, up 32% year-on-year and 1% sequentially, as more OEMs adopt Nvidia’s end-to-end self-driving stack.

Margins were another area of strength. Non-GAAP gross margin expanded sequentially to 73.6% from 72.7%, even as Nvidia transitions from the older Hopper HGX architecture to the more complex, higher-performance Blackwell systems. Year-on-year margins dipped slightly due to mix shift toward full-scale data center solutions, but the sequential improvement is what mattered most for investors—and it came exactly as analysts hoped.

Operating expenses moved meaningfully higher as expected, rising 38% year-on-year on a non-GAAP basis and 11% sequentially, reflecting increased compute infrastructure costs, higher compensation tied to workforce growth, and heavy R&D spending to support next-generation Blackwell and Rubin platforms. Nvidia is spending aggressively because demand requires it; the pipeline remains so large that the company must continue building out compute infrastructure and securing long lead-time components.

Other income provided another lift. Interest income reached $624 million, while non-marketable and publicly-held equity gains drove overall OI&E to $1.4 billion. Cash on the balance sheet rose to $60.6 billion from $38.5 billion a year ago, supported by record operating cash flow of $23.8 billion, up sharply from both the prior quarter and prior year. Nvidia returned $12.7 billion to shareholders through buybacks and dividends during the quarter.

One powerful data point buried in the details: multi-year cloud service agreements jumped from $12.6 billion last quarter to $26.0 billion. That doubling in a single quarter signals a dramatic expansion in Nvidia’s enterprise cloud footprint and supports the notion that demand visibility extends well into 2026 and beyond.

And then came the outlook—arguably the most important piece of all. Nvidia guided to $65 billion in revenue for Q4, ±2%. That’s not just ahead of consensus near $61–62 billion—it’s a full-blown acceleration from Q3’s record haul. Non-GAAP gross margin is expected around 75.0%, with GAAP margin close behind at 74.8%. If that holds, Nvidia would be exiting the year at the high end of the “mid-70s” gross margin trajectory analysts hoped for, despite persistent concerns about memory costs, liquid cooling supply, and power-related bottlenecks across CSPs and OEMs.

Operating expenses are expected to rise again to $5.0 billion non-GAAP, consistent with Nvidia’s strategy of front-loading investment during this unprecedented demand cycle. Even so, the implied operating leverage remains extraordinary—this company is expanding expenses at a fraction of its revenue growth rate.

With these numbers, Jensen Huang will almost certainly deliver an upbeat narrative on the call, likely reinforcing themes from GTC: a $3–4 trillion AI infrastructure opportunity this decade, more than $500 billion in cumulative Blackwell and Rubin orders through 2026, and the ramp from 20 million combined GPUs expected to ship across 2025 and 2026. Any confirmation of additional CoWoS capacity, Blackwell Ultra deployments, or sovereign AI initiatives will only strengthen the long-term case.

For now, Nvidia gave investors everything they asked for: a decisive revenue beat, margin stability, monster data center growth, and a Q4 guide that resets expectations even higher. The only remaining test is whether the market rewards it—or whether a sell-the-news reaction risks becoming the most ironic plot twist of the AI supercycle. But on fundamentals alone, Nvidia just delivered a quarter that will be remembered.