NVIDIA says demand for its AI infrastructure could exceed available production capacity by roughly $100bn next year, while the company is also using guarantees and other financial support to help customers buy more of its hardware. The result is a revealing picture of an AI boom constrained not by appetite, but by chips, memory, power and money.
NVIDIA Watch analysis
What happened
In an analysis published by CRN, NVIDIA executives said the company’s suppliers are struggling to keep pace with demand. CFO Colette Kress said annual revenue could potentially double next year without supply constraints, while the company expects growth of about 70 per cent under current conditions. The figures come from NVIDIA’s own earnings commentary and CRN’s reporting.
The pressure is spreading beyond GPUs. CRN cites shortages of memory, SSDs, CPUs and power-supply components, with some channel partners reporting delivery times of eight to 12 months. Memory prices are also rising sharply, according to Kress, creating the pleasingly circular problem of AI infrastructure demand helping cause the shortage that limits more AI infrastructure.
NVIDIA is also helping customers secure financing. CRN reports that the company has agreed to guarantee up to $105bn for a data centre leased to OpenAI, with payments triggered if OpenAI defaults. It may provide as much as $125bn in residual-value support for projects with Goldman Sachs, Blackstone and other partners planning to fund AI infrastructure purchases.
Why it matters
This is not only a chip-supply story. NVIDIA’s financial backing can help AI labs, neoclouds and data-centre operators obtain hardware they might otherwise struggle to finance. That may accelerate deployment, but it also ties the supplier more closely to the success of the customers buying its products.
The arrangement deserves attention because NVIDIA is both selling into the infrastructure rush and helping keep that rush funded. The company says these measures respond to genuine customer demand rather than creating it. Critics, including investor Michael Burry, have described the activity as circular financing. The supplied evidence does not settle that dispute.
Our read
The useful takeaway is that AI’s next bottleneck may be industrial rather than imaginative. The market can dream up more models quickly. Manufacturing, memory, electricity, data-centre space and credit are less easily persuaded to sprint.
NVIDIA’s guarantees may be commercially rational, but they make the AI build-out harder to read from headline sales alone. When the hardware supplier is also helping customers buy the hardware, investors should examine who carries the downside if demand cools or projects slip. The engine is powerful. It is also financing part of its own fuel bill.
What to watch
- Whether NVIDIA’s projected $100bn supply gap narrows as memory and system capacity expand.
- Whether customers receive the promised hardware on schedule or continue facing year-long delays.
- How much of NVIDIA’s financial support is ultimately drawn, and by whom.
- Whether regulators and investors demand clearer disclosure of guarantees and residual-value commitments.
Discussion spark: Should NVIDIA’s customer guarantees be treated as sensible infrastructure finance, or as a warning that AI demand is being amplified by the companies selling the hardware?
Sources and evidence
- Analysis: What Nvidia’s Massive Supply-Demand Gap Says About AI Mania – crn.com (17 September 2026, 21:12 UTC)
Independent WittyWires tracker for public updates about NVIDIA. Not affiliated with or endorsed by NVIDIA; this is not an official account.