Nvidia Q2 FY27 Results: The $91 Billion Question, Answered

Nvidia delivered a beat so wide it reset the debate. Second-quarter fiscal 2027 revenue came in at $96.22 billion on August 26 — more than $4 billion above the $92.17 billion Wall Street consensus and nearly $5 billion clear of the company's own $91 billion guidance. Adjusted earnings per share hit $2.22 versus the roughly $2.09 analysts expected, GAAP earnings per diluted share reached $2.46, and Data Center revenue surged to $89.0 billion, up 117 percent year over year. Then management raised the bar again: Q3 FY27 revenue guidance landed at $108 billion, plus or minus 2 percent.

Here's how the numbers shook out — and why this print reads less like an earnings beat and more like a statement about where the AI economy is headed.

The results vs. expectations

What were Nvidia's Q2 FY2027 results? Nvidia's Q2 FY2027 results are a revenue of $96.22 billion, up 18% sequentially and 106% year over year — a 4% beat against the $92.17 billion Wall Street expected. Data Center generated $89.0 billion, which is a full 92% of the total. According to Nvidia's report, GAAP net income reached $59.7 billion, up 126% year over year, and GAAP earnings per share came in at $2.22 versus $2.09 expected. For example, this marks the third consecutive quarter in which Nvidia added roughly $15 billion of new quarterly revenue. In practice, the clearest signal in the release is gross margin: GAAP gross margin held at 75.0%, which shows pricing power intact despite the HBM memory cost squeeze. First, Data Center grew 117% year over year; second, networking attach keeps climbing; finally, the guide assumes zero China compute revenue — the upside lever if export policy loosens.

The streak is getting hard to ignore. In Q4 fiscal 2026 Nvidia posted record revenue of $68.1 billion; in Q1 fiscal 2027 it blew past that with $81.6 billion while raising its dividend 25-fold and announcing an $80 billion buyback. Q2 FY27 extended the pattern: revenue of $96,221 million was up 18% sequentially and 106% year over year, with GAAP net income of $59.7 billion, up 126 percent from a year ago. Gross margin held at 75.0 percent on both GAAP and non-GAAP bases — evidence that the pricing power hasn't cracked.

Every headline metric topped expectations. Capital returns kept pace too: Nvidia repurchased shares and paid dividends totaling approximately $26.0 billion during the quarter, leaving roughly $99.0 billion under its repurchase authorization. The next quarterly dividend of $0.25 per share is payable October 1, 2026, to shareholders of record on September 10.

What drove the quarter

Jensen Huang framed the moment in typically grand terms. "AI has reached its inflection point," he said in the release. "It's doing useful work. Its tokens are productive and profitable. Now, compute is revenue." A year ago, he noted, a single lab was driving the infrastructure buildout; today he points to multiple frontier labs scaling in parallel, a flourishing ecosystem of AI labs and startups, a thriving open-model ecosystem, and physical AI coming online.

The product cadence backed him up. The Vera Rubin platform is now in full production, with racks already running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius — ahead of the September ramp that analysts had penciled in. Spectrum-6 switch systems, supporting both pluggable and co-packaged optics, are arriving across the world's gigascale AI factories, and Nvidia also unveiled Vera, a CPU built specifically for AI agents.

On margins, the guidance-era math held: roughly 75 percent gross margin reflects the premium attached to being the default supplier of AI compute, with software and networking attached to every GPU sale. The longer-standing risk — margin compression as custom silicon from Anthropic and hyperscalers scales — remains the thing to watch, but this quarter offered no sign of it.

The three shifts that made this quarter possible

• Multiple frontier labs are spending simultaneously. A year ago, one major lab drove the bulk of infrastructure buildout. Today several labs scale in parallel, open-model ecosystems are thriving, and AI startups are placing their own hardware orders — the demand base has diversified, making the revenue stream more durable than it was even two quarters ago.
• Physical AI is no longer theoretical. Huang's line that "compute is revenue" is worth unpacking: workloads have expanded beyond chatbots and code assistants into robotics, autonomous systems, and industrial applications. These tend to be compute-intensive and latency-sensitive — playing directly to Nvidia's accelerated-computing strengths and opening spending categories that weren't part of the AI capex conversation 18 months ago.
• The networking attach rate keeps climbing — Spectrum-6 switch systems with pluggable and co-packaged optics ship to gigascale AI factories globally. Every rack sold carries networking and software revenue alongside the GPUs: a full-stack structural advantage custom-silicon rivals have struggled to replicate.

Q3 guidance is the real headline

What is Nvidia's Q3 FY2027 guidance? Nvidia's Q3 FY2027 guidance is a revenue outlook of $108.0 billion, plus or minus 2% — roughly 12% above the record Q2 and above the ~$100 billion analysts projected before the print. According to Nvidia's CFO commentary, the outlook assumes zero Data Center compute revenue from China and GAAP gross margins of 74.0%. In simple terms, Nvidia is guiding 12% sequential growth while writing off one of the world's largest AI hardware markets — that means any export-policy loosening flows straight to upside that current guidance does not reflect. Data shows the guide implies adding $11.8 billion of new quarterly revenue, in line with the recent run-rate; studies indicate that guidance beats of this size have historically preceded further upside revisions. First, diversified demand: multiple frontier labs are spending simultaneously. Second, physical AI: robotics and autonomous workloads expand the addressable spend. Finally, full-stack attach: networking and software revenue ride along with every rack.

As always with Nvidia, the forward number mattered more than the reported one — and it did not disappoint. Q3 FY27 revenue is guided to $108.0 billion, plus or minus 2 percent, well above the ~$100 billion mark bulls had floated before the print, and roughly 12 percent above the record just set. Two details stand out:

• No China assumption. The outlook includes zero Data Center compute revenue from China. Growth of this size is being produced despite a geopolitical handbrake, not because of friendly conditions.
• Margins tick down slightly. Guidance calls for GAAP and non-GAAP gross margins of 74.0 percent, plus or minus 50 basis points — modest normalization as new platforms ramp, not distress.

For the full fiscal year 2027, Nvidia expects GAAP and non-GAAP tax rates between 16.0 and 18.0 percent.

What the bears still get right

No honest read of this quarter ignores the risks. Custom silicon is real: Google's TPUs are mature, Amazon's Trainium scales, and Broadcom builds accelerators for multiple hyperscalers — none has dented Nvidia's margins yet, but the cumulative effect over 12–24 months could shift the math, particularly for inference workloads where Nvidia's training dominance is less of a moat. Memory remains a bottleneck: HBM3e pricing and the coming HBM4 transition feed directly into Nvidia's bill of materials while SK Hynix and Samsung ramp and demand still outstrips supply. China is a wildcard: the zero-revenue assumption is conservative, but export controls could tighten further — or diplomatic shifts could open partial access. And valuation assumes flawless execution — at current multiples, a guide that merely meets consensus could trigger a significant repricing, whatever the order book says.

Nvidia Stock Forecast 2026: What the Q2 Beat Means for NVDA

Where is Nvidia stock headed after the Q2 FY27 beat? The post-earnings picture is this: NVDA trades near $205–$230 in September 2026, while the mean 12-month analyst target is a $324 consensus across 55 analysts, with the high at $515 and post-beat initiations at $345. According to MarketBeat, that spread reflects a market pricing an accelerating infrastructure cycle against a macro overhang — the September 30 Fed decision. For example, Barchart's post-print target of $345 implies roughly 60% upside from the trading band. In practice, the cleanest framing is a three-checkpoint watchlist: first, the Fed's September 30 decision, which moves multiple sentiment; second, Rubin ramp yield commentary in the November Q3 report; finally, whether China export language shifts from zero assumed to anything permissive. Studies indicate that post-earnings target revisions of this scale tend to cluster before the next print, which means the window between now and November is when the street's view consolidates.

The Q2 beat ($96.22B revenue vs $92.17B expected, Data Center $89B, Q3 guidance $108B) strengthens the bull case for NVDA through 2026, but the stock’s path now depends on two variables: the September 30 Fed decision and China data-center revenue assumptions baked into guidance. Analysts tracking the print note that guidance quality — not the beat itself — is what separated winners from losers this earnings season. For a stock priced for perfection, executing a $108B quarter with zero China revenue assumed is the kind of upside optionality forecast models rarely price in.

The post-print trajectory: what fresh targets say

Since the August 26 print, the setup has sharpened rather than cooled. NVDA has oscillated between roughly $205 and $230 — while the 55-analyst mean target sits near $324 and post-beat initiations landed at $345 (MarketBeat; Barchart). That gap is the market pricing two things at once: an infrastructure cycle still accelerating, and a macro overhang — the September 30 Fed decision — that has nothing to do with Nvidia's order book. Huang's own framing has escalated accordingly: not just beating quarters, but $1 trillion in annual revenue by 2027, with Groq-based inference racks and the Vera CPU extending the platform beyond GPUs (Data Center Knowledge; CNBC).

Where does that leave the stock? The bull case writes itself from the order book: diversified customers, an early Rubin ramp, zero-China upside, and inference demand the Groq deal now addresses directly. The bear case remains what it was — custom silicon maturing, HBM costs, and a valuation that assumes flawless execution. The difference after this print: the burden of proof has moved to the bears. Watch three checkpoints through year-end — the Fed's September 30 decision, Rubin yield commentary in the Q3 report, and whether China export language shifts from "zero assumed" to anything more permissive.

For the memory-side bottleneck behind this story, see our Samsung 400+ layer BV-NAND explainer; for the infrastructure buildout context, the $500B AI infrastructure push and the $3B Ohio AI campus power play cover the physical layer; and for the open-source angle, our Hugging Face deal analysis connects the dots.

Frequently Asked Questions

When did Nvidia report Q2 FY27 earnings?

Nvidia reported its Q2 FY2027 results on August 26, 2026. Revenue came in at $96.22 billion versus the $92.17 billion analysts expected. The full figures are available on Nvidia's official newsroom.

What were Nvidia's Q2 FY27 results?

Q2 FY27 revenue was $96.22 billion against $92.17 billion expected, with adjusted EPS of $2.22. Data Center revenue surged to about $89 billion, and Nvidia returned roughly $26.0 billion to shareholders through buybacks and dividends during the quarter.

What is Nvidia's Q3 FY27 guidance?

Nvidia guided Q3 FY27 revenue to approximately $108 billion, plus or minus 2 percent — with zero Data Center compute revenue from China assumed and gross margins guided to 74.0 percent. The guidance reset expectations for the AI infrastructure buildout more than the Q2 beat itself.

Where is NVDA stock headed after the Q2 FY27 beat?

The stock has traded between roughly $205 and $230 since the print, while the 55-analyst mean twelve-month target sits near $324 and post-beat initiations reached $345. The path from here depends on three things: the September 30 Fed decision, Rubin ramp yields in the Q3 report, and any shift in China export policy — levers that cut both ways.

Why is the China revenue assumption important?

The $108 billion Q3 outlook includes zero Data Center compute revenue from China. Guiding 12 percent sequential growth while writing off one of the world’s largest AI hardware markets means any export-policy loosening flows straight to upside that current guidance does not reflect — and a further tightening would remove an assumption that is already at zero.

The bottom line

The $91 billion question that framed the preview was answered emphatically: at $96.22 billion, Nvidia didn't just clear its bar — it moved it again, to $108 billion for Q3. The competitive questions aren't gone; custom chips, memory costs, and China policy still hang over the story. But this report made one thing plain: demand for AI compute is outrunning supply, and the industry's gauge reading says the boom still has headroom.