Nvidia Corporation — Key Risks
Export Controls Are Cutting NVIDIA Off From Major Markets — Especially China
The U.S. government has imposed a series of escalating restrictions on NVIDIA's ability to sell its most powerful chips abroad. NVIDIA explicitly states it is "effectively foreclosed from competing in China's data center computing/compute market" as of fiscal year 2026. The H20 chip restrictions alone triggered a $4.5 billion charge in Q1 FY2026 for excess inventory and purchase obligations. With China historically a large market and new rules potentially expanding worldwide, this risk is ongoing and unpredictable.
A Handful of Customers Account for a Disproportionate Share of Revenue
In fiscal year 2026, one customer represented 22% of total revenue and another represented 14% — meaning just two buyers drove over a third of all sales. Most orders are placed on a purchase-order basis, meaning customers can reduce or cancel commitments with little notice. Losing or being blocked from selling to even one of these customers could meaningfully hurt results.
Demand Forecasting Is Extremely Difficult, Creating Costly Supply Mismatches
NVIDIA builds products months in advance based on demand estimates, but those estimates are routinely wrong — in both directions. When they overestimate, they face inventory write-downs and cancellation penalties; when they underestimate, they miss revenue. The company is now accelerating its product launch cadence (aiming for a new data center architecture annually), which makes this problem harder. Gross margins in Q2 FY2025 were already hurt by inventory provisions for low-yielding Blackwell material.
NVIDIA Depends Almost Entirely on Outside Manufacturers It Does Not Control
NVIDIA designs chips but relies on a small number of third-party foundries (primarily TSMC) and contract manufacturers for all production, assembly, and testing. It cannot guarantee supply, quality, or delivery schedules. Geographic concentration — particularly in Taiwan and South Korea — means geopolitical disruption or natural disaster in those regions could halt production. Lead times can already exceed 12 months.
Antitrust Regulators Worldwide Are Scrutinizing NVIDIA's Market Position
Regulators in the EU, U.S., UK, China, South Korea, and Japan are actively requesting information about NVIDIA's GPU sales, supply allocation, and AI partnerships. China's antitrust authority issued a preliminary finding in September 2025 that NVIDIA violated terms of its Mellanox acquisition approval. Potential outcomes include financial penalties, restrictions on its networking business, or forced changes to business practices.
Product Defects Can Be Costly and Hard to Catch Before Shipment
NVIDIA's hardware and software are deeply complex, and defects — including in AI models trained on third-party data — have occurred in the past and may again. Because its chips are embedded in large data center systems, a single flaw can cascade into expensive recalls, warranty costs, inventory write-offs, and damaged customer relationships. The filing notes that some failures are only discovered after a product has shipped.
A Key Licensing Bet on Groq's Technology May Not Pay Off
NVIDIA made a significant, nonrefundable payment to license technology from Groq, Inc., and must now invest substantial engineering effort to incorporate it into future products. There is no guarantee the technology will work as planned, be adopted by customers, or generate enough revenue to recover the cost — and delays are explicitly acknowledged as possible.