Hyperscaler Capex Surge Reaches $725B as NVIDIA and Wall Street Launch $500B Compute Financing Platforms
The capital expenditure cycle behind artificial intelligence infrastructure has reached unprecedented heights in the second quarter of 2026, with the top U.S. hyperscalers—Amazon, Alphabet, Microsoft, Meta, and Oracle—on track to spend a combined $725 billion on capex over the next 12 to 18 months, with a staggering 75% to 80% of those budgets directly allocated to AI-specific compute, networking, and specialized data center infrastructure.
To fund this historic investment cycle without overwhelming corporate balance sheets, a paradigm shift in AI infrastructure financing has emerged. On August 10, 2026, NVIDIA Corporation (NVDA) and six of the world's premier financial institutions officially launched a series of strategic partnerships to mobilize over $500 billion in third-party capital, turning high-performance compute into a new, financeable institutional asset class.
The $500 Billion Compute Financing Platform as an Asset Class
NVIDIA signed Memorandums of Understanding (MOUs) with Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR to establish independent compute financing platforms. This initiative is designed to treat AI compute infrastructure much like commercial real estate, toll roads, or other long-duration, cash-flow-generating assets to borrow against.
By using private credit, insurance funds, and institutional capital to underwrite GPUs and data centers, NVIDIA is helping its end users—including leading frontier AI labs, neoclouds, enterprises, and AI-native startups—secure financing at attractive rates without tapping their own balance sheets. NVIDIA CEO Jensen Huang framed this milestone as the creation of a new asset class:
"We began by building chips; today, we are helping create a new class of productive, investable infrastructure: AI factories. In AI, compute is revenue... These financing platforms will help customers access scarce compute at scale and build the DSX AI factories that will power every industry and country in the age of AI."
Jon Gray, President and COO of Blackstone, added on CNBC that AI compute will be treated as a "financeable asset class" in the same way that mortgage lenders look at homes, allowing institutional capital to directly fund the backbone of the global digital economy.
NVIDIA's Credit Support and Gross Guarantee Disclosures
While Wall Street provides the long-duration capital, NVIDIA is actively using its own massive cash pile to provide credit support, guaranteeing a portion of these transactions to satisfy underwriting banks. In its Q2 FY2027 earnings report (released August 26, 2026), NVIDIA disclosed a maximum gross guarantee exposure of $108.5 billion.
This exposure is highly concentrated in strategic ecosystem partnerships:
- $105.0 billion in credit support and guarantees is dedicated to SB Energy's Ohio tech campus, which will host NVIDIA compute leased directly to OpenAI (this was previously rumored in the market to be as high as $250 billion, but was finalized at $105 billion).
- $3.5 billion backs lease obligations for other AI cloud partners (neoclouds).
- $50.0 billion represents direct equity and debt investments in frontier AI labs.
The "Circular Financing" Debate
This extensive financial plumbing has drawn intense scrutiny from Wall Street critics, who argue that NVIDIA is engaging in "circular financing"—essentially investing in and guaranteeing leases for its own customers to artificially inflate demand for its chips. NVIDIA CFO Colette Kress directly addressed these concerns during the Q2 FY2027 earnings call:
"We recognize the scale of this support, and we know some will call this circular financing. We see it differently. We believe these investments, measured against the strength of their demand, the business they create for us, the ecosystem they build on Nvidia's platform, and the equity returns on our invested capital will be excellent, and our risk is limited."
Kress explained that for smaller, AI-native neoclouds, NVIDIA guarantees a minimum portion of a data center's capacity. In exchange, NVIDIA takes a cut of the provider's rental revenue above that threshold, effectively getting paid twice—once on the initial hardware sale, and again through a share of the rental revenue.