Uh Oh… Nvidia’s $100 Billion Deal With OpenAI Has Fallen Apart

OpenAI appears to have stepped back from a previously reported $100 billion letter-of-intent to buy Nvidia compute, citing concerns that Nvidia’s latest chips are inadequate for inference workloads; after months of negotiation the deal stalled and Bloomberg later reported Nvidia may instead invest about $20 billion in OpenAI. The change in plans coincided with OpenAI signing deals with rivals such as AMD and contributed to Nvidia shares dropping roughly 9% over five days (over 7% in the past month), though both Jensen Huang and Sam Altman publicly denied any lasting strain. The episode underscores investor unease about massive AI infrastructure spending and could temper near-term capex commitments and valuation momentum for AI hardware leaders.

Analysis

Market structure: The unraveling of a $100B LOI and the reported OpenAI pivot to AMD shifts near-term winner/loser dynamics toward AMD (positive) and away from NVDA (negative), as shown by NVDA’s ~7% month-to-date and ~9% five-day declines. Inference demand is carving out a lower-margin, high-volume segment where price/performance and deployment costs (power, HBM memory) matter more than raw training throughput, eroding some of NVDA’s pricing power if large buyers diversify.

Risk assessment: Tail risks include OpenAI/mega-clouds moving material share to AMD or custom accelerators (20–40% revenue risk to NVDA in a multi-year worst-case), regulatory/export controls on advanced nodes, and capital-allocation missteps from large equity stakes. Immediate (days) risk = elevated IV and headline-driven moves; short-term (weeks–months) = deal confirmations and guidance revisions; long-term (quarters–years) = structural split between training (NVDA-dominant) and inference (more fragmented).

Trade implications: Tactical trades should exploit volatility and directional exposure—hedge NVDA equity exposure with 3-month puts or buy AMD stock for market-share upside; consider pair trades (long AMD, short NVDA) sized to net neutral dollar exposure. Options strategies (buy 90–120 day 20% OTM NVDA puts or sell NVDA 30–60 day covered calls where IV>40%) capture asymmetric risk; rotate 1–3% portfolio weight into cloud software beneficiaries (MSFT/AMZN) that monetize inference.

Contrarian angle: Consensus may overstate permanent damage to NVDA; training demand and ecosystem lock-in (CUDA, software stack) keep NVDA advantaged for years—a sell-off of >20% from current levels would likely be a buying opportunity. Historical parallels (Intel/ARM competitive cycles) show hardware leadership can reassert after short-term supplier diversification; watch implied vol thresholds (NVDA IV>50%) as trigger for premium-selling or opportunistic re-entry.

Key Decisions for Investors

• Establish a 1–1.5% portfolio hedge against NVDA (ticker NVDA) by buying 3-month puts ~20% OTM within the next 7–14 days; size to cover 50% of existing NVDA exposure or to represent a 1% portfolio downside hedge, close if NVDA falls >25% or IV spikes >70%.
• Open a 2–3% long position in AMD (ticker AMD) funded by trimming NVDA exposure by an equal dollar amount; target 3–6 month hold with an outperformance goal of 20–30% vs NVDA, cut AMD at -15% absolute or if NVDA/AMD spread reverses by 10% in favor of NVDA.
• Implement a 30–60 day income overlay on existing NVDA positions by selling covered calls at ~+10% strikes when NVDA 30-day IV >40%; collect premium and re-evaluate on earnings or material OpenAI announcements.
• Allocate 1–2% into cloud/AI software beneficiaries (MSFT or AMZN) over 1–12 months to capture services/inference monetization (buy MSFT or AMZN, hold 6–12 months), increase allocation by +1% if OpenAI confirms multi-year AMD deal >$5B.
• Trigger-based action: if within 30–90 days OpenAI publicly signs a multi-year inference contract >=$5B with AMD or another rival, reduce NVDA exposure by an additional 5–10% and re-allocate to AMD/MSFT; conversely, if NVDA re-enters a financing round >=$10B or OpenAI reaffirms exclusivity within 60 days, add to NVDA up to +3% portfolio on dips <=15%.