Nvidia, Meta Have $300 Billion AI Problem : US Pioneer Global VC DIFCHQ SFO NYC Singapore – Riyadh Swiss Our Mind

This article first appeared on GuruFocus.

Big Tech’s artificial-intelligence boom is creating a new financial risk investors may not see clearly on corporate balance sheets. Nvidia (NASDAQ:NVDA), Meta Platforms (NASDAQ:META), Broadcom (NASDAQ:AVGO) and other technology companies have provided as much as $300 billion in guarantees supporting AI data centers and chips over the past year, allowing the infrastructure buildout to accelerate while leaving much of the potential exposure outside conventional debt.

NVDA GF Value chart

The structures typically guarantee a minimum future value for data centers or computing equipment owned by special-purpose entities. That makes projects easier and cheaper to finance, but the technology company can become responsible for part of the shortfall if those assets eventually fetch less than the guaranteed amount.

The scale is becoming significant. Nvidia has agreed to provide as much as $105 billion of support for SB Energy’s Ohio data-center campus serving OpenAI, while also backing as much as 25% of certain other financing packages using Nvidia hardware.

Broadcom has maximum exposure of about $29 billion from backing AI-rack leases connected to Anthropic. Meta, meanwhile, supplied a $28 billion guarantee supporting its Hyperion data-center venture with Blue Owl, helping the project raise $27 billion while keeping most of the obligation outside Meta’s reported balance sheet.

Morgan Stanley estimates seven major cloud and chip companies have accumulated more than $3.1 trillion in off-balance-sheet commitments and other credit support, a broader category than the $300 billion of guarantees themselves.

The structure works particularly well while AI demand remains strong. The risk appears if computing capacity becomes oversupplied, customers struggle financially or rapidly improving chips cause older hardware to depreciate faster than expected.

Investor Takeaway

Investors should increasingly look beyond headline capex and reported debt when assessing the AI buildout. Maximum guarantee exposure, customer credit quality and the residual value of AI equipment now matter alongside revenue growth. Nvidia’s $105 billion commitment is especially important because its eventual cost depends partly on OpenAI continuing to meet long-term obligations. Upcoming quarterly filings from Nvidia, Broadcom and Meta should be watched for expanding guarantees, changes in estimated liability values and signs that financing structures are growing faster than the cash flows they are designed to support.

https://finance.yahoo.com/technology/ai/articles/nvidia-meta-300-billion-ai-145204122.html?shem=aimgspc,