First reported 2026-11-17 by NVIDIA fiscal Q3 FY2027 earnings call / CFO commentary
NVIDIA Locks In $390B Supply Commitments, Mostly Memory — Margin Guide Cut on HBM Costs
NVIDIA's supply and capacity commitments jumped to $390 billion from $279 billion in a single quarter — an increase management attributed mainly to multiyear memory purchases — with $203 billion of that falling in fiscal 2028 versus $87 billion disclosed three months earlier. The memory cost pass-through is already visible in guidance: Q3 gross margin of 74.8% is guided down to 71–72% in Q4, which management called the bottom before a 72–73% fiscal 2028 recovery, while CSIS notes SK hynix is sold out of DRAM and HBM through 2029.
A ~3-point gross-margin guide-down explicitly blamed on memory confirms HBM/DRAM is now the binding BOM cost line for accelerators, not logic wafers or packaging. Multiyear prepay-style commitments of this size effectively pre-empt 2027–2028 HBM supply for NVIDIA and will push second-tier accelerator and AI-server builders into worse pricing and longer lead times.
Chips affected — model the economics
This signal touches these AI accelerators. Open any in the cost calculator, pre-filled with its full BOM.
Sources
- Yahoo Finance — Nvidia Added $111 Billion of Supply Commitments in a Quarter, Mostly Memory. Micron Is Still Priced Like a Bust
- Global Data Center Hub — NVIDIA Q3 FY2027 Earnings: OpenAI and PORTS-Pike Guarantees
- CSIS — Between a Rock and a Hard Place: Evaluating the Risks of Chinese Memory Chips
- Yahoo Finance — Memory supply constraints will be even 'tighter' in 2027: Micron CTO
Related data & tools
Silicon Analysts Research
All benchmarks and data are derived from publicly available sources (earnings calls, press releases, analyst reports, regulatory filings). Figures are estimates for educational purposes only and should not be used as the sole basis for business or investment decisions. Public sources only — no insider data. Generated from the cited reports; verify against the originals before acting. Full Terms & Data Provenance