Nvidia's $750 Billion AI Bet: Inside the SK Group Partnership and the OpenAI Guarantee

The $500 Billion+ SK Group Partnership

On July 24, 2026, Nvidia and SK Group signed letters of intent for a comprehensive AI infrastructure partnership valued at more than $500 billion . The agreement has three main components:

• AI Factory: SK Telecom will build a 2-gigawatt AI data center powered by Nvidia's Vera Rubin DSX platform, expected to come online by 2027 .
• Memory Supply: SK Hynix, which controls roughly 60% of the global high-bandwidth memory (HBM) market, is locked in as Nvidia's long-term HBM supplier and will co-develop next-generation HBM .
• Naver Investment: Nvidia committed approximately $1 billion to the Korean search giant Naver as part of the broader AI ecosystem buildout in South Korea .

The deal was announced at a San Francisco AI Summit alongside South Korean President Lee Jae Myung, with Nvidia CEO Jensen Huang calling it "business partnerships that will represent over $500 billion of business together" .

The ~$250 Billion OpenAI Guarantee

On July 26, the Wall Street Journal reported that Nvidia is in early-stage talks to provide roughly $250 billion in financing guarantees for OpenAI . The guarantees would help OpenAI lease computing capacity from a 10-gigawatt data center campus that SoftBank's SB Energy is developing in Piketon, Ohio .

How it would work: Nvidia would backstop OpenAI's lease payments, allowing the ChatGPT maker to secure more favorable financing terms than conventional lenders would offer given its lack of investment-grade credit . Separately, Nvidia is discussing $350 billion in additional financing for GPU purchases by OpenAI, bringing the total potential Nvidia-backed exposure to OpenAI to roughly $600 billion . The full Ohio project could cost more than $500 billion to build, including the chips inside .

This deal structure is the third time in under a year Nvidia has restructured how it finances its biggest customer, and each iteration has drawn sharper scrutiny .

Critics' Three Main Concerns

1. Circular Financing

The core concern is a self-reinforcing loop: Nvidia provides financing or guarantees to customers (OpenAI, SK Group), who then use that money to buy Nvidia GPUs. This means Nvidia essentially books revenue that it funded itself .

Goldman Sachs analyst James Schneider warned as early as October 2025 that Nvidia's equity financing to customers like OpenAI was starting to resemble "circular revenue" — investments flow out as funding, returns come back as chip sales . Short sellers Jim Chanos and Michael Burry have made similar arguments, despite Nvidia issuing a seven-page memo denying that it engages in vendor financing .

Some analysts say this artificially inflates both Nvidia's revenue and perceived AI demand . As one Bloomberg analysis put it: "What looks like organic growth may be, at least in part, an artifact of the financing structure" .

2. Unprecedented Scale of Financial Risk

Nvidia's market capitalization is roughly $3 trillion and its annual free cash flow was approximately $97 billion in fiscal 2026 . The combined ~$750 billion in potential obligations would be roughly 8x annual free cash flow and a significant share of the company's entire market value .

Critics argue this is dot-com era level leverage, concentrated in a single counterparty (OpenAI) and one unproven infrastructure buildout . Nvidia's shares fell 4–5% on the news amid these fears .

3. Gap Between Cash Flows and Backed Obligations

A financial guarantee of $250 billion means Nvidia would be on the hook if OpenAI cannot pay its lease—yet OpenAI is still a private company with no public balance sheet and a history of massive operating losses . The gap between Nvidia's $97 billion in annual FCF and the $250+ billion in contingent liabilities under discussion is the central source of investor alarm .

As one analysis put it, Nvidia would effectively be "co-signing a lease" for a customer that conventional lenders won't touch .