Six major financiers signed on to fund Nvidia's AI build-out. Buried in the deal: Nvidia's promise to cover up to 25% of losses if the hardware loses value, a guarantee that could hit $125 billion, with much of that risk headed for insurers' balance sheets. On August 10, Nvidia announced memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to build financing platforms meant to mobilize more than $500 billion in third-party capital for AI infrastructure, according to Axios.
The pitch to these firms was straightforward: raise money from pension funds, sovereign wealth funds and insurers, use it to buy Nvidia chips and build data centers, then lease the capacity to AI companies for years of steady payments. To make that pitch land with investors who don't normally gamble on chip depreciation, Nvidia added a sweetener. It said it "may provide a residual-value support mechanism for up to 25% of an opportunity, assessed carefully on a project-by-project basis." That covers part of the gap if hardware turns out to be worth less than expected when a lease ends.
Do the math across every deal Nvidia is courting, and that backstop could total $125 billion in exposure, per Axios's reporting on the plan. That's Nvidia, the company selling the shovels in this gold rush, agreeing to personally guarantee that the shovels won't rust. Here's the part that should catch your attention: none of that $500 billion sits on Nvidia's own balance sheet, and neither does most of the credit risk behind it.
The GPUs get bought by special purpose vehicles. The vehicles issue debt. And increasingly, according to CNBC's reporting from April 2026, that debt gets placed with insurance and retirement capital managed by firms like Apollo and KKR.
Data centers are expensive, decades-spanning projects, and they happen to match the long-dated liabilities insurers already carry - things like annuities and pension obligations. The risk doesn't disappear. It just changes owners, and it arrives on an insurer's books relabeled as investment-grade fixed income.
CNBC's April reporting quotes the phrase "GPU debt treadmill," coined by AI commentator Dave Friedman, to describe the core mismatch driving this whole structure. GPUs have a useful life of roughly seven years. The data centers housing them are built to last 20 to 30.
Lenders and insurers underwriting a 20-year facility are implicitly betting that today's Nvidia chips, or their replacements, will still be worth something in year fifteen, even though the hardware itself will have been swapped out two or three times by then. Capacity is also a live problem. One insurance executive told CNBC that insuring a single $20 billion data center campus was nearly impossible to price in 2023.
By 2026, it's become, in that executive's words, a weekly conversation. Insurers have responded with new products built for this exposure: cover for credit losses, cover for declines in chip resale value, and cover for contract breaches tied to power outages or cooling failures at the facilities themselves, according to the same CNBC report. The scale of the debt underneath all of this has grown fast.
Incremental borrowing funded about 9% of hyperscaler capital spending in fiscal 2024. By mid-2026 that figure had climbed to roughly 32% on a trailing basis, and U.S. data-center debt issuance roughly doubled to about $182 billion in 2025, per the reporting Axios cited in its financing coverage. Frankly, that's the kind of growth curve that makes credit analysts nervous even when the underlying business is healthy.
Why the timing matters Nvidia isn't doing this in a vacuum. Bain & Company said this week that the AI industry needs to generate $6 trillion a year in new revenue by 2031 just to justify the data center spending already committed. Existing AI services might cover $1.8 trillion of that, leaving a $4.2 trillion gap to fill from businesses that barely exist yet, like autonomous robotics and AI-driven drug discovery.
Bain separately projects $5 trillion to $6.5 trillion in data center spending through 2030. Those are the numbers Wall Street is holding up against Nvidia's financing push. They explain why insurers are being asked to carry so much of the weight: the traditional lenders and Nvidia's own balance sheet can't absorb spending at that scale alone.
Nvidia's arrangement doesn't eliminate risk. It reroutes it, from Nvidia's income statement and from the banks that might otherwise hold this debt, into the reserves of companies whose entire business model depends on correctly pricing tail risk decades in advance. If AI demand keeps compounding the way Nvidia's revenue guidance assumes, the residual-value guarantee mostly sits unused, a marketing device that made institutional capital comfortable enough to write the check.
But if a major AI lab pulls back capacity, or a cheaper chip generation arrives faster than expected and strands the current fleet, the 25% backstop gets tested for real. Insurers holding the debt behind it find out how much of that "investment-grade fixed income" was actually GPU depreciation risk wearing a different label. None of the six firms named in the MOUs have disclosed how much capital they've actually committed, or whether the residual-value terms are even finalized.
Nvidia's own release says the partnerships remain subject to final agreements. That's worth sitting with: the market has already reacted to $500 billion in financing infrastructure and a $125 billion guarantee ceiling, and so far, not a single dollar of it is confirmed as spent. Also read: TSMC is raising 3nm chip prices up to 15% and nobody can say no • Meta's Muse AI promises to run your life but it just gave away a stranger's address • OpenAI apologizes after its AI agent hacked Australia's Medicare system in June This article is posted in AI News , check it out for more related stories.
Source: Startup Fortune
Texas · Hometown Post

