NVDA — Q4 FY26 Earnings Review (GauchoRico)

Date: 2026-03-01 Quarter: Q4 FY26 (three months ended January 25, 2026; reported February 25, 2026) Position: 26.6% (7.12% LEAPS, 19.48% shares) — largest position Prior thesis: Intact — sole arms dealer for the AI buildout

Headline

Holy smokes!! $68.1 billion in a single quarter. Let me say that again — sixty-eight billion dollars in ninety days. That is a $272 billion annualized revenue run-rate for what was a $27 billion company three years ago. Revenue growth re-accelerated to +73% YoY (from +63% last quarter), gross margins expanded back to 75%, and free cash flow hit $34.9 billion — more FCF in one quarter than most companies generate in a decade. And then they guided Q1 FY27 to $78 billion, implying another re-acceleration to +77% YoY. At this scale, this kind of acceleration is unprecedented in the history of technology companies. This was not just a good quarter — this was one of the most dominant earnings reports I have ever seen from any company.

The Numbers — 12 Quarters

| | Q1_FY24 | Q2_FY24 | Q3_FY24 | Q4_FY24 | Q1_FY25 | Q2_FY25 | Q3_FY25 | Q4_FY25 | Q1_FY26 | Q2_FY26 | Q3_FY26 | Q4_FY26 | | | Apr-23 | Jul-23 | Oct-23 | Jan-24 | Apr-24 | Jul-24 | Oct-24 | Jan-25 | Apr-25 | Jul-25 | Oct-25 | Jan-26 | |---|---|---|---|---|---|---|---|---|---|---|---|---| | Revenue ($B) | 7.2 | 13.5 | 18.1 | 22.1 | 26.0 | 30.0 | 35.1 | 39.3 | 44.1 | 46.7 | 57.0 | 68.1 | | YoY % | -13.2% | +101.5% | +205.5% | +265.3% | +262.1% | +122.4% | +93.6% | +77.9% | +69.2% | +55.6% | +62.5% | +73.2% | | QoQ % | +18.9% | +87.8% | +34.2% | +22.0% | +17.8% | +15.3% | +16.8% | +12.1% | +12.0% | +6.1% | +22.0% | +19.5% | | GM [GAAP] | 64.6% | 70.1% | 74.0% | 76.0% | 78.4% | 75.1% | 74.6% | 73.0% | 60.5% | 72.4% | 73.4% | 75.0% | | GM [Non-GAAP] | — | — | 75.0% | 76.7% | 78.9% | 75.7% | 75.0% | 73.5% | 61.0% | 72.7% | 73.6% | 75.2% | | Op Margin [GAAP] | 29.8% | 50.3% | 57.5% | 61.6% | 64.9% | 62.1% | 62.3% | 61.1% | 49.1% | 60.8% | 63.2% | 65.0% | | Net Margin [GAAP] | 28.4% | 45.8% | 51.0% | 55.6% | 57.1% | 55.3% | 55.0% | 56.2% | 42.6% | 56.5% | 56.0% | 63.1% | | EPS [GAAP] | $0.82 | $0.25 | $0.37 | $0.49 | $0.60 | $0.67 | $0.78 | $0.89 | $0.76 | $1.08 | 1.30|**1.76** | | EPS [Non-GAAP] | — | — | $0.40 | $0.52 | $0.61 | $0.68 | $0.81 | $0.89 | $0.81 | $1.05 | 1.30|**1.62** | | FCF ($B) | 2.6 | 6.0 | 7.0 | 11.2 | 14.9 | 13.5 | 16.8 | 15.5 | 26.1 | 13.5 | 22.1 | 34.9 | | FCF % | 36.7% | 44.8% | 38.8% | 50.9% | 57.3% | 44.9% | 47.8% | 39.5% | 59.3% | 28.8% | 38.7% | 51.2% |

TTM (Q1-Q4 FY26): Revenue $215.9B | FCF $96.6B | Non-GAAP EPS $4.77

Revenue Assessment

Revenue came in at $68.127B, beating the $65.0B midpoint guide by $3.1B (+4.8%). This was the fourth consecutive beat-and-raise in FY26 with an average beat of ~4.2%. At this scale, beating guide by nearly 3.2billionisreally, reallyimpressive.Toputitinperspective, NVDA^′sQ4FY26beatalone(3.1B) is larger than many growth companies' entire quarterly revenue.

The trajectory is what gets me most excited. Look at the YoY growth pattern:

Growth decelerated into Q2 (the Blackwell supply bottleneck) and then re-accelerated for three consecutive quarters. A company doing $68B/quarter re-accelerating from 56% to 73% growth — that's pretty incredible. And the Q1 FY27 guide of $78B implies ~+77% YoY (vs. Q1 FY26 $44.1B), meaning management sees the acceleration continuing. That is extraordinarily rare at this revenue scale.

Data Center Deep Dive

Data Center revenue was $62.3B, +75% YoY, +22% QoQ. That's 91.5% of total revenue — the highest DC mix we've seen. Within Data Center:

The networking number is staggering. $11 billion in a single quarter, up 263% YoY. That's driven by NVLink fabric for GB200/GB300 systems. This is NVIDIA's moat expanding — they are not just selling you GPUs, they are selling you the entire interconnect fabric. The more GPUs customers buy, the more networking revenue follows. It's a compounding flywheel.

Hyperscaler revenue was "slightly over 50% of Data Center revenue" — so roughly 31 − 32Bfromhyperscalersalone.Butcrucially, growthwas * ledbynon − hyperscalercustomers * .Thatmeanstherevenuebaseisdiversifying.SovereignAInowrunsat>30B annually (>3x YoY growth), and I count Canada, France, Netherlands, Singapore, and the UK as active sovereign AI customers. Enterprise, telco, and sovereign — all pulling.

Margins — The Story Gets Better

Gross margins expanded to 75.0% GAAP / 75.2% Non-GAAP, up from 73.4% / 73.6% last quarter. This is important because the bear case was that Blackwell systems-level selling (NVL72 racks with cooling, networking, etc.) would structurally compress margins. Instead, Colette Kress said margins improved because of "improved mix and cost structure" as Blackwell continued to ramp. Blackwell is getting cheaper to produce as yields and volumes improve. The CFO explicitly said she expects "mid-70s" gross margins long-term.

Operating margins hit 65.0% GAAP / 67.7% Non-GAAP — best in the cycle. Operating expenses grew 45% YoY (GAAP), which sounds like a lot but revenue grew 73%. That's significant operating leverage at scale.

FCF of $34.9B is a 51.2% FCF margin. In a single quarter, NVIDIA generated more free cash flow than the entire annual FCF of all but maybe 4-5 companies on Earth. For the full year FY26, FCF was $96.6B. They returned 41.1Btoshareholders(40.1B buybacks + $974M dividends) and still ended the year with $62.6B in cash and marketable securities. The balance sheet is a fortress.

Management Commentary & Forward View

Jensen framed the quarter around the "agentic AI inflection point." Key quotes:

• "Grace Blackwell with NVLink is the king of inference today — delivering an order-of-magnitude lower cost per token."
• "Vera Rubin samples shipped to customers this week" — on track for H2 CY2026 production.
• Blackwell Ultra: "up to 50x better performance and 35x lower cost" for agentic AI.
• Rubin platform: "up to a 10x reduction in inference token cost" vs. Blackwell.

The roadmap density here is important. NVIDIA is shipping Blackwell now, Blackwell Ultra later in CY2026, and Vera Rubin in H2 CY2026. Each generation delivers a massive step-function improvement. The pace of innovation makes it nearly impossible for competitors to catch up — by the time AMD or custom silicon gets close to Blackwell, NVIDIA is already shipping Rubin with a 10x cost-per-token improvement.

Q1 FY27 Guidance: $78.0B ±2%

This implies +77% YoY growth and +14.5% QoQ sequential growth. No China Data Center compute revenue assumed — zero. That means if H200 exports materialize, it's upside to the number. Gross margin guidance of 75.0% Non-GAAP ±50bps confirms the "mid-70s" commitment is holding.

One important policy change: starting Q1 FY27, SBC will be included in Non-GAAP measures. This is actually a positive signal from management — it means they're treating SBC as a real expense, not sweeping it under the rug. The impact is $1.9B per quarter (0.1% on GM, $1.9B in OpEx). Historical Non-GAAP comparisons will need adjustment.

I give management an A+ for execution. They beat revenue by $3.1B, expanded margins, guided to a re-acceleration, and delivered the Blackwell ramp exactly as promised.

Beat Pattern & Guidance Credibility

The average beat is ~4.2% above midpoint, and the pattern has been consistent all year. If you apply that same 4.2% beat to Q1 FY27 guidance of 78B, youget 81.3B. At $81B, that's 84% YoY growth. At a $325B annualized run-rate.

That sounds crazy. But then again, everything about NVIDIA for the past two years has sounded crazy — and the numbers keep getting more crazy, not less.

Inventory & Supply

Inventory: $21.4B (+8% QoQ). Total supply commitments: $95.2B. That's $95 billion in forward commitments to suppliers — NVIDIA is buying every TSMC CoWoS wafer and HBM chip they can get their hands on. This is not a company worried about demand softening. This is a company that can't build enough product fast enough.

Cash and marketable securities: $62.6B. Even with $95B in supply commitments, they have the balance sheet to back it up many times over. DSO came down from 53 to 51 days — customers are paying faster, not slower.

Share count: 24,430M, down from 24,706M a year ago. Buybacks are reducing the float by about 1.1% annually. At a $3.4T market cap, even $40B in annual buybacks only moves the needle modestly, but it's accretive and disciplined.

Risks I'm Watching

1. FY27 gross margin sustainability. Mid-70s commitment held this quarter and into Q1 guide. But HBM pricing, TSMC packaging costs, and NVLink switch complexity are all real headwinds. If margins settle at 72-73% instead of 75%, that's $5-7B less annual gross profit at $300B+ revenue. I think management can hold it — their leverage is innovation speed (each new generation delivers better performance-per-watt, which is the "lever" Colette described) — but it's the thing to watch.

2. Custom silicon competition. Google TPU v7 Ironwood, Amazon Trainium2, Meta MTIA are all real. But Jensen's framing is smart — custom silicon for inference doesn't displace NVIDIA from training. And now Anthropic has committed 1GW to NVIDIA (their first time on NVIDIA hardware). The same customers use both. I've said before: NVIDIA is innovating too fast for anyone to catch within 3 years. This quarter reinforces that view.

3. Hyperscaler CapEx cyclicality. $600B estimated CY2026 CapEx is an enormous number. If these hyperscalers pull back in 2027-28, it matters. The counter: inference demand from reasoning models (thinking tokens), agentic AI, and physical AI scales compute demand structurally. More efficiency → more usage → more compute. Jensen's Jevons paradox framing is correct in my view.

4. China. Zero DC compute revenue from China assumed in Q1 FY27 guide. H200 approved but not in outlook. This is actually de-risking — if China opens up, it's pure upside.

Valuation

Valuation is rich on absolute terms but increasingly reasonable relative to growth. A PEG ratio of 0.4x for the most dominant technology franchise on Earth is not expensive. The stock has been roughly flat since October 2024 while earnings have grown 80%+. This is classic multiple compression — the market is discounting a slowdown that has yet to materialize. At some point, if NVIDIA keeps delivering 60-80% growth with 50%+ FCF margins, the stock re-rates. Or it just compounds at the earnings growth rate while the multiple stabilizes. Either way, I'm happy holding.

Call me crazy, but I think NVDA shares could double two more times from here in the next 5-6 years. $68B/quarter growing at 70%+ with 75% gross margins and 50% FCF margins — there isn't another company like this on the planet.

Action

Hold. No change to allocation.

NVDA remains my largest position at 26.6%. The quarter gave me zero reasons to trim. Growth re-accelerated, margins expanded, guidance implies continued acceleration, and the competitive moat is widening not narrowing. The networking revenue explosion ($11B, +263% YoY) is particularly bullish — it means NVIDIA owns more of the total data center stack than ever before.

I'm not adding here because (a) the position is already 26.6% which is enormous, and (b) valuation, while reasonable on PEG, is not a fat-pitch dip where I'd want LEAPS exposure. If we get a 15-20% pullback on macro noise or some rotation event, I'd consider adding via LEAPS calls. But absent that, this is a hold-and-let-it-compound position.

Thesis status: Intact — strengthening. Every metric that matters improved this quarter. The sole arms dealer for the AI buildout continues to be the only game in town.

We will see what happens...

GR

Sources: NVIDIA CFO Commentary EX-99.2 (Q4 FY26), NVIDIA Newsroom press release (Feb 25, 2026), SEC EDGAR XBRL, investing.db, Atlas Q3 FY26 baseline analysis.