Nvidia's $750B+ AI Deals Spark Circular Financing Concerns
Taylor Wilson
Nvidia is advancing AI infrastructure deals worth over $750 billion in total, but critics warn that the chip maker's practice of financing its own customers — who then spend that money on Nvidia chips — is artificially inflating industry demand and valuations.
How big are these deals?
The largest sits in South Korea: Nvidia and SK Group will co-build over 2 GW of AI data centers worth more than $50 billion. The first "AI factory," built by SK Telecom, is expected to go live next year.
CEO Jensen Huang called it "a golden age for Korea." SK Group's subsidiary SK Hynix is one of the world's largest AI memory-chip suppliers. This means → the deal ties chip maker, memory supplier, and data-center builder onto a single chain.
A more complex deal is still in early talks: Nvidia may guarantee up to $25 billion to help OpenAI lease a $50 billion, 10 GW data-center hub that SoftBank is developing in Ohio, while separately discussing $35 billion in financing for OpenAI's chip purchases.
What exactly is "circular financing"?
In plain terms = Nvidia lends money or provides guarantees to a customer; the customer uses that capital to buy Nvidia's chips; chip revenue flows back to Nvidia. The money circulates inside the system, creating a self-reinforcing loop.
Allspring portfolio manager Gary Tan noted: "Nvidia's investments reinforce confidence in long-term AI build-out, but capital is increasingly being used to fund future AI customers."
Global X strategist Billy Leung added that guaranteeing OpenAI's data-center debt "deepens an already-scrutinized vendor-financing arrangement" — a demand signal and a funding-pressure signal at the same time.
Is Nvidia the only one doing this?
No. Google has pledged to backstop Anthropic's lease payments across five data centers, helping the OpenAI rival secure loans equivalent to $35 billion.
This reflects a broader pattern: the AI industry is entering a "big companies guaranteeing each other" financing model, with chip makers, cloud providers, and AI firms financially intertwined.
In plain terms = if any link breaks — say AI demand falls short — the chain reaction travels along the guarantee chain. The industry's systemic risk exposure is rising.
Why hasn't the stock kept up?
Despite a steady stream of deal announcements, Nvidia shares have gained only about 11% year-to-date, while the Philadelphia Semiconductor Index is up 67% over the same period.
Pre-market Monday, Nvidia rose 1.1% to $209.16 — but Intel, Broadcom, and AMD gained 2.8%, 2.6%, and 3.2% respectively, all outpacing Nvidia.
This means → the market has no quarrel with Nvidia's earnings power, but it is discounting the sustainability of a financing-driven growth model — voting with its feet.
How concentrated is SoftBank's bet?
SoftBank's cumulative committed investment in OpenAI has reached nearly $65 billion, plus a $40 billion bridge loan — one of the largest bridge financings in Asia-Pacific history.
OpenAI remains unlisted and unprofitable. SoftBank's heavy reliance on a single loss-making company is stoking investor unease.
In plain terms = whether AI demand ultimately materializes to support this debt chain is the make-or-break variable for the industry's valuation logic. If it doesn't, circular financing flips from "growth flywheel" to "debt vortex."
Content is for reference only, not financial advice.