Will NVIDIA's non-GAAP gross margin fall below 70% in any quarter by Q2 FY2027?

Q4 FY2026 non-GAAP gross margin of 75.2% decisively resolves one of the three resolution windows well above the 70% threshold. Q1 FY2027 is guided at 75.0% +/-50bps, establishing a floor of 74.5% at the low end of guidance — still 4.5 points above 70%. This leaves only Q2 FY2027 as the genuinely uncertain quarter. The prior context identified three risk vectors: rising input costs, architecture transition margin compression, and ASIC pricing pressure. Q4 data addressed all three favorably — margins expanded despite rising costs, Blackwell ramp is mature without compression, and no competitive pricing concessions are visible. The only plausible path to sub-70% is a one-time inventory write-down analogous to Q1 FY2026's H20 charge, which would require a new export control action. The Rubin transition begins H2 CY2026 but Vera Rubin samples already shipping suggests smoother overlap. A 5+ point buffer with only 2 remaining quarters makes sub-70% a genuine tail event.

The margin trajectory through FY2026 tells a clear expansion story: 61.0% (Q1 with charge) → 72.7% (Q2) → 73.6% (Q3) → 75.2% (Q4). The exit rate is at the top of the 'mid-seventies' guidance range. However, I weight slightly higher than consensus because the committee's unresolved debate on COMPETITIVE_POSITION (DEFENSIBLE vs DOMINANT) has direct margin implications. If DEFENSIBLE is correct and ASIC pricing pressure intensifies through CY2026, there could be a more aggressive pricing response from NVIDIA to defend market share in the Q2 FY2027 timeframe. Additionally, the Rubin transition introduces product-mix uncertainty — GB300 to Rubin Ultra pricing and margin profiles are not yet disclosed. The SBC methodology change (0.1% impact) is negligible but represents an additional margin headwind. Still, a 5+ point compression from 75% to below 70% in a single quarter without a one-time charge is historically unprecedented for NVIDIA.

The mathematical case against sub-70% is now very strong. Q4 FY2026 at 75.2% is the highest non-GAAP GM in FY2026. Q1 FY2027 at 75.0% guided represents management's high-confidence near-term outlook. For Q2 FY2027 to breach 70%, margins would need to decline approximately 500 basis points in a single quarter. The only precedent for such a decline is Q1 FY2026's H20 write-down, which was a $4.5B charge triggered by export control enforcement. No analogous charge is identifiable: China DC compute revenue is zero in Q1 FY2027 guidance, eliminating the primary write-down risk vector. The Groq acquisition ($13B) could introduce some integration-related costs but these would likely flow through operating expenses, not COGS. NVLink Fusion's co-opting strategy suggests NVIDIA is maintaining pricing power through ecosystem integration rather than discounting. The networking segment at $11.0B (+263% YoY) is a higher-margin mix component that continues growing fastest.

With Q4 FY2026 reporting 75.2% non-GAAP GM, one of three resolution windows is decisively eliminated. Q1 FY2027 guided at 75.0% +/-50bps makes a sub-70% outcome in that quarter essentially impossible barring an extraordinary mid-quarter event. This concentrates all remaining risk into Q2 FY2027, which won't report until ~August 2026. The margin expansion from 73.6% to 75.2% despite acknowledged rising input costs demonstrates that pricing power is intact. The CFO attributed improvement to 'improved mix and cost structure' — Blackwell's maturation is margin-accretive, not dilutive. For Q2 FY2027 specifically, the Rubin transition is the primary risk vector, but Vera Rubin samples shipping in February 2026 with production in H2 CY2026 suggests production overlap rather than a gap. The committee found no evidence of margin-damaging competitive pricing in the Q4 data.

The probability should be adjusted significantly downward from the prior 14%. The 75.2% Q4 result exceeds expectations and eliminates one window. The 75.0% Q1 FY2027 guide with +/-50bps range means even the worst case (74.5%) has a 450 basis point buffer. For Q2 FY2027, I need a scenario where margins drop 500+ bps from the guided run-rate. The prior analysis identified the H20 write-down as the only precedent, and that required an exogenous policy shock plus an already-existing inventory of restricted chips. No such inventory exists currently — China DC compute is guided at $0 for Q1 FY2027. ASIC shipments projected +44.6% in 2026 vs +16.1% for GPUs, but NVIDIA's margins expanded even as ASIC share grew, suggesting NVIDIA is capturing the highest-margin segments of the market. The 'mid-seventies' full-year guidance implies management expects Q2 FY2027 to remain in the 74-76% range.

I assign a slightly higher probability than some peers because I want to account for genuine uncertainty about Q2 FY2027. The resolution window extends through September 2026, and Q2 FY2027 (ending July 2026) has several months of unknown variables. The Rubin architecture transition creates legitimate uncertainty about product mix and margin profiles. Additionally, geopolitical risk cannot be dismissed — a significant expansion of export controls to Gulf states or allies could trigger inventory write-downs or demand disruption that compresses margins. The 'mid-seventies' guidance is management's best estimate but NVIDIA has missed margin guidance before (Q1 FY2026 was guided at 71% non-GAAP and came in at 61% including the charge). However, the structural setup — 75.2% exit rate, expanding trajectory, Blackwell maturation, networking mix shift — makes a sub-70% outcome require multiple simultaneous negative events.

Q4 FY2026 at 75.2% eliminates one resolution window. Q1 FY2027 guided at 75.0% +/-50bps makes sub-70% essentially impossible. Only Q2 FY2027 remains as a risk quarter. The margin trajectory is expanding (73.6% → 75.2%), and management guides 'mid-seventies' for the full year. With a 5+ point buffer and only one uncertain quarter, the probability is very low. The only realistic scenario is a large one-time inventory charge, and no such charge vector is apparent in current disclosures.

The updated data is unambiguously favorable for a NO resolution. Margins expanded 160bps QoQ to 75.2% in Q4. Networking revenue at $11.0B (+263% YoY) is the highest-margin segment and is growing fastest. Blackwell is now mature at 2/3 of DC revenue with no margin dilution. Q1 guided at 75.0%. The mathematical path to sub-70% requires either a massive one-time charge (no identified vector) or a simultaneous collapse of pricing power, input cost spike, and product mix degradation in Q2 FY2027. This confluence is extremely improbable.

Strong consensus toward lower probability given Q4 FY2026 results. The 75.2% actual and 75.0% Q1 FY2027 guidance provide substantial buffer. I assign slightly higher than my model peers to account for the possibility that an exogenous shock — new export control expansion, sudden competitive pricing war, or a manufacturing issue in the Rubin ramp — could create an unexpected margin event in Q2 FY2027. The non-GAAP methodology change (SBC inclusion) adds a negligible 10bps headwind. Overall, sub-70% is now a low-probability tail event requiring exceptional circumstances.