Jensen Huang Said Nvidia's $30 Billion OpenAI Investment Might Be the Last. Nvidia Just Guaranteed Up to $105 Billion of OpenAI's Leases.

Nvidia disclosed residual value guarantees for OpenAI data-center leases in Ohio (PORTS-Pike) capped at $105B, with payment only triggered if OpenAI becomes insolvent/defaults or stops paying rent—otherwise Nvidia delivers nothing today. The structure is designed to terminate early if OpenAI achieves a satisfactory credit rating, and the campus is targeted to start commissioning in 2028. The article frames this as manageable given Nvidia’s scale (Q1 FY2027 revenue $81.6B, +85% YoY) alongside strong guidance (Q2 revenue guidance ~$91B), suggesting limited near-term earnings risk despite added linkage to OpenAI.

Analysis

This is less a credit story than a distribution-channel story: NVDA is turning scarce AI supply into a financed ecosystem, which raises switching costs and makes future compute procurement harder for rivals to dislodge. The immediate beneficiary is NVDA’s full-stack franchise; the less obvious loser is any GPU/vendor that competes on product alone, because the bar is now product + capital support. Over time, that should help preserve NVDA’s pricing power and attach rates in networking/software, not just accelerators.

The market’s real risk is not the headline liability size but the regime in which it matters. If AI capex cools and customer funding tightens, the contingent support could become visible exactly when NVDA growth is already decelerating, creating a double hit to sentiment and multiple. That makes this a 6-18 month valuation overhang rather than a near-term cash-flow issue; the first hard catalyst is the next earnings guide and any language around future commitments, not the lease dates themselves.

Contrarian view: bears may be overreading balance-sheet exposure while bulls may be underreading the strategic optionality of vendor-financed demand. The structure effectively monetizes NVDA’s fortress equity value to lock future revenue with very little current cash drag. The thesis breaks if NVDA’s guide slips materially below Street or if similar guarantees start to proliferate without corresponding share gains, signaling the market is forcing financing concessions to sustain demand.

Key Decisions for Investors

• Stay long NVDA into the next 1-3 month earnings/capex cycle, but add only on post-print weakness; thesis is invalidated if revenue guide or gross margin shows any sequential deceleration versus current expectations.
• Pair trade: long NVDA / short AMD for the next 1-3 months. NVDA is increasingly monetizing ecosystem lock-in plus financing optionality; AMD lacks that balance-sheet lever. Cover if AMD wins a large named AI design win or NVDA commentary turns defensive.
• Do not short SFTBY or data-center proxies on this headline alone; the economics are too indirect. Put them on watch for 6-12 months and only act if project-level funding stress shows up in disclosure or financing spreads widen.
• If NVDA sells off >5% on "contingent liability" fear without a guide cut, consider a call-spread structure rather than outright stock; the market may be mispricing a non-cash, back-ended risk.