NVIDIA Q4 FY2026 Earnings: We Called $78B, Rubin Is Ahead of Schedule

NVIDIA reported Q4 FY2026 earnings on February 25, 2026 — and the results confirmed what our supply-chain model projected. NVIDIA Q4 FY2026 earnings results showed the company guiding Q1 FY2027 revenue to $78 billion, landing at the top of our pre-earnings $73–78B estimate and $7 billion above the Street's $71B consensus. The stock dropped 5.5% the next day anyway. For institutional investors tracking NVIDIA earnings, the gap between the fundamental signal and the price action is the entire story.

Three days before the report, we published our earnings preview built on TSMC CoWoS capacity data, HBM allocation analysis, and hyperscaler capex commitments. That framework projected $73–78B. NVIDIA delivered $78B. This post unpacks what the earnings call revealed about Rubin's timeline, why the stock sold off despite a massive beat, and where NVDA goes from here — with specific price corridors for the next few days, next month, and next few months.

The Setup: Our Pre-Earnings Call

On February 24 — one day before NVIDIA reported — we published a detailed earnings preview built on supply-chain triangulation rather than consensus-tracking. The core thesis: the Street's Q1 FY2027 guidance estimate of ~$71 billion was too low. Our bottom-up math, derived from TSMC CoWoS capacity expansion (~26% sequential increase to ~70K KWPM), NVIDIA's ~69% allocation share, and $660–690 billion in committed hyperscaler capex, pointed to $73–78 billion.

We also flagged the sell-the-news risk explicitly. NVIDIA had sold off on 5 of the last 7 earnings beats. Polymarket was pricing a 94.5% probability of a beat. The options market implied ±8–10% movement. Our scenario framework assigned a 50% probability to a "beat + in-line guide" outcome yielding flat to +3%, and a 30% probability to a "strong guide" outcome ($75B+) yielding +7–10%. The actual result — a massive beat with a 5.5% selloff — fell into a pattern we explicitly warned about: the market gives no credit for durability.

The guidance landed in our strong-guide scenario. The stock reaction didn't. That divergence is the subject of this post.

What Actually Happened

NVIDIA delivered Q4 FY2026 results that beat on every metric that matters:

The headline: $78 billion in Q1 FY2027 guidance. That's not just above consensus — it's $7 billion above consensus, a 10% upside surprise on the single most important number in the report. Data center revenue of $57.1 billion grew 11.5% sequentially and represented 85% of total revenue. Blackwell architecture GPUs drove the acceleration, with GB300 systems comprising roughly two-thirds of data center shipments.

Jensen Huang's commentary was unambiguously bullish: demand for Blackwell is "incredible," the $500 billion backlog is not contracting, and lead times remain at 40–52 weeks versus a normal 12–16 weeks. The company also disclosed that inference workloads now represent a "rapidly growing" share of data center revenue — a qualitative signal that the platform re-rating thesis is building, even if NVIDIA hasn't yet quantified the exact percentage.

Why the Stock Dropped 5.5% Despite a Beat

NVIDIA guided $78 billion — $7 billion above consensus — and the stock dropped 5.5% the next day. For anyone not tracking the pattern, this looks irrational. It isn't. Three forces explain the reaction.

1. The sell-the-news pattern is structural, not anomalous. NVIDIA has now sold off in 4 of the last 8 post-earnings sessions despite beating consensus in every single one. The pattern is clear: as beat magnitude moderates (from 11%+ surprises in FY2024 to 4–7% in FY2026), the stock's willingness to reward beats diminishes. The market has learned to expect the beat. What moves the stock is the second derivative — whether the beat is accelerating or decelerating.

2. The beat was fully priced in before the report. Polymarket had a 94.5% probability of a beat. Goldman Sachs was modeling a ~$2B revenue beat. The options market was pricing ±8–10% movement with a bullish skew (put/call volume ratio of 0.41). When 94.5% of the market expects a beat and the beat arrives, the information content of the event is near zero. The only way to generate a positive reaction was to beat the beat expectations — and while $78B guidance was strong, gross margin guidance of ~75% (flat sequentially) gave bears a talking point.

3. Gross margin flatness spooked the growth-at-any-price crowd. Non-GAAP gross margins came in at 75.2% for Q4 and were guided to ~75% for Q1 FY2027. That's not bad — it's excellent for a company growing revenue 50%+ annually. But the market wanted to see margin expansion alongside revenue expansion. Flat margins at $78B revenue implies the revenue growth is coming from volume, not pricing power — and volume-driven growth is what semiconductor companies do. Pricing-power-driven growth is what platform companies do. The margin signal cut against the re-rating narrative.

Rubin Status: Ahead of Schedule

The most important qualitative signal from the earnings call wasn't the guidance number — it was Jensen Huang's language on Rubin. Mapping his exact words against our prediction framework from the pre-earnings preview reveals that Rubin is ahead of where the Street expected it to be.

The key phrase: "Rubin samples have been shipped to customers." This is ahead of the consensus timeline, which had samples arriving in Q2–Q3 2026. Shipping samples in late February means NVIDIA is running approximately one quarter ahead of schedule on the Blackwell-to-Rubin transition — the exact transition that bears had flagged as a potential "air pocket" risk.

The customer commitments are equally significant. Meta and Anthropic have already committed to Rubin-based infrastructure. Meta's commitment is particularly notable given its $65 billion 2026 capex budget and its history of being NVIDIA's largest single customer. Anthropic's commitment — alongside its $10 billion investment from NVIDIA — signals that the next generation of frontier AI training will run on Rubin silicon.

The 8 Supply Chain Signals to Monitor

Our investment framework tracks 8 supplier-level signals that provide 12–24 months of forward visibility into NVIDIA's revenue trajectory. Post-earnings, all 8 remain bullish. Here's the current status of each:

The most important signal is #8 — the earnings call language itself. When a company with $500 billion in backlog and 40–52 week lead times tells you demand is "incredible" and lead times are not contracting despite massive capacity expansion, that's the supply-chain equivalent of a flashing green light. Demand is growing faster than supply can expand. That's the single most bullish data point in the entire semiconductor channel.

Signal #6 (Samsung HBM3e) remains the one to watch most closely. SK Hynix currently supplies ~90% of NVIDIA's HBM, but that share is expected to drop to ~50% as Samsung ramps. If Samsung's yields don't improve on schedule, HBM becomes the new bottleneck just as CoWoS loosens — a constraint substitution that could cap revenue growth even with abundant packaging capacity.

Forward Outlook: Days, Weeks, Months

Post-earnings price action in mega-cap semiconductors follows a predictable three-phase pattern: the initial reaction (1–3 days), the catalyst-driven re-pricing (2–4 weeks), and the fundamental convergence (1–3 months). Here's our framework for each phase.

Next few days ($185–200 corridor). The 5.5% selloff took NVDA from ~$195 pre-earnings to ~$184 at the close on February 26. Historical patterns suggest the initial selloff overshoots: in 6 of the last 8 quarters, the stock recovered 30–50% of the initial post-earnings move within 3 trading days. That puts the near-term range at $185–200, with the $190 max-pain level (195,208 contracts at the $200 March strike) acting as a gravitational center. Short-term traders will be watching the $185 support level — a break below it opens $175–180.

Next month ($185–205, GTC catalyst). GTC on March 17 is the next binary event. If Jensen demos working Rubin silicon and provides an updated roadmap showing Rubin production ahead of schedule, the stock re-rates toward $200–205 as the Blackwell-to-Rubin transition risk collapses. If GTC disappoints — no silicon demo, vague timeline language — the stock likely retests $180–185 as the market prices in transition uncertainty. The options market will begin pricing GTC risk approximately one week before the event; watch for IV expansion in March 21 expiry options as a leading indicator.

Next few months ($195–230 by June). The fundamental case is straightforward: NVIDIA just guided $78 billion for Q1 FY2027. If they deliver — and the supply-chain data says they will — the FY2027 revenue estimate moves from ~$335 billion to $340–350 billion. At 30–35x forward earnings, that supports a $210–235 stock price. The base case of $195–215 assumes the market continues to apply a "prove it again" discount and waits for Q1 results (late May) before re-rating. The bear case ($175–190) requires a macro shock, a hyperscaler capex pause, or a Samsung HBM failure that constrains supply.

Key Risks to the Thesis

Every bullish thesis has failure modes. Here are the four that matter most right now:

1. Flat gross margins signal volume, not pricing power. Non-GAAP gross margins at ~75% are excellent in absolute terms but flat sequentially. The market wants to see margin expansion as evidence that NVIDIA is a platform company (pricing power) rather than a semiconductor company (volume-driven). If margins stay flat or compress through FY2027, the "platform re-rating" thesis weakens and the stock trades at semiconductor multiples (25–30x) rather than platform multiples (35–45x). The difference is $50–80 per share.

2. China revenue remains approximately zero. The H200 export ban has effectively zeroed out NVIDIA's China data center revenue. Before the ban, China represented 20–25% of data center sales. That's $10–15 billion in annualized revenue that doesn't exist in current estimates. Any signal of resumed sales through compliant products would be pure upside — but the geopolitical trajectory suggests further restrictions are more likely than relaxation. The risk is that competitors (Huawei Ascend 910C) fill the vacuum permanently.

3. Gaming supply constraints are a margin drag. Gaming revenue was $3.4 billion in Q4, up 11% YoY but constrained by supply allocation decisions that prioritize data center GPUs. If NVIDIA continues to starve the gaming channel to feed data center demand, it risks losing mindshare in a segment that historically provided high-margin, counter-cyclical revenue. The RTX 5090 launch has been supply-constrained since day one — that's a deliberate choice, not a manufacturing problem.

4. The $10 billion Anthropic investment creates perception risk. NVIDIA's $10 billion investment in Anthropic is strategically sound (securing a major Rubin customer) but optically problematic. Bears will argue it's a form of vendor financing — NVIDIA investing in a customer to ensure that customer buys NVIDIA products. The historical parallel is Cisco in 1999–2000, which financed customers who subsequently defaulted. The situations are not identical (Anthropic has real revenue and Amazon backing), but the narrative risk is real and will surface in every bear thesis for the next 12 months.