Overview
On August 10, NVIDIA announced MOUs with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish independent AI compute infrastructure financing platforms, designed to mobilize over $500 billion of third-party capital "over the coming years." Official announcement
A key detail: the $500B is the total third-party capital the platforms are designed to mobilize—not NVIDIA revenue, not a single fund, and not a commitment to any one customer.
An Open Compute Capital Market
This is a structural innovation. Previously, AI infrastructure financing was project-by-project (each data center or GPU cluster financed separately). NVIDIA now wants to make AI factories into an investable asset class:
- Customers: frontier AI labs, AI-native startups, enterprises, cloud providers, national AI services
- Capital: independent dedicated pools from the six financial institutions
- Risk: each deal independently underwritten on customer, demand, utilization, cash flow, and residual value
- NVIDIA's role: provides the AI factory platform (DSX architecture) and may offer up to 25% residual value support on select projects
- Wide adoption: CUDA ecosystem across clouds, OEMs, enterprises
- Software upgrades: each software generation improves performance, efficiency, and TCO of installed infrastructure
- Hardware longevity: A100 (2020) still in commercial use in 2026
- Cross-customer fungibility: one AI factory serves multiple customers and workloads
- H100 one-year rent: ~$1.70/GPU-hour (Oct 2025) → ~$2.35/GPU-hour (Mar 2026), +38%
- Cross-cloud on-demand median: ~$2.00 (Oct 2025) → ~$2.70 (Jun 2026), +35%
- Blackwell B200 premium: $5.30–7.05/GPU-hour
- NVIDIA's 5-year CDS rose ~5.3 bp to 77.215 bp—the largest single-day increase in two weeks
- Apollo rose 3.59%, Blackstone rose 3.3%—markets saw the financial institutions' roles as safer than NVIDIA's own
- Sep 2025: up to $100B investment in OpenAI (conditional on OpenAI deploying at least 10 GW of NVIDIA systems)
- Jan 2026: $2B investment in CoreWeave to help build 5+ GW of AI factories by 2030
- xAI data center deal: participation as anchor LP in a $5.4B structured deal
- Actual deployment pace: how many billions of real transactions materialize?
- CDS spread trajectory: back below 70 bp, or holding above 80 bp?
- AI factory REITs/SPVs: will dedicated securitization vehicles for retail allocation emerge?
- The narrative battle: whether "circular financing" becomes the AI bull market's next major risk story
- The $500B is a design target, not a commitment; actual deployment is unknown
- Terms of NVIDIA's ≤25% residual value support are not public
- Interest rates, tenors, and collateral structures per deal are undisclosed
- DSX AI factory architecture specifics (GPU config, network topology, power) are not public
- Capital pool allocations among the six partners are undisclosed and may be uneven
- https://nvidianews.nvidia.com/news/nvidia-partners-with-apollo-blackrock-blackstone-brookfield-goldman-sachs-and-kkr-to-establish-ai-compute-infrastructure-financing-platforms-to-mobilize-over-500-billion-of-third-party-capital
- https://x.com/JensenHuang/status/2086934705207959965
- https://www.fortuneindia.com/technology/nvidia-goes-beyond-chips-teams-up-with-wall-street-for-a-500-billion-ai-buildout/153198
Jensen Huang's Four Q&As
Q1: Is this circular financing? "This initiative is designed in response to those concerns. We're bringing independent, long-term institutional capital into AI infrastructure. Demand is real. Capital partners underwrite each project independently."
Q2: Why does NVIDIA support financing? "In some cases NVIDIA may provide up to 25% residual value support, evaluated project by project. This support is limited, residual-based, and designed to complement—not replace—independent underwriting. It is significantly lower than other compute financing arrangements."
Q3: Can the market absorb this capacity? "The question isn't whether we're building data centers. It's whether we're building productive AI factories. Capacity will be built around real customer economics."
Q4: Where's the return? "Returns lie in AI's usefulness. More compute creates better AI; better AI creates more usage; more usage creates more revenue; more revenue drives more compute."
The Investability Argument
NVIDIA argues its compute is a durable asset:
Rental price evidence:
CDS Up 5.3 Basis Points: Markets Not Fully Convinced
On the day after the announcement, NVIDIA shares fell 2.86% to $217.55 (market cap ~$5.3T). More telling:
The CDS move reflects implicit doubt about the "$500B circular financing" model. Meanwhile, Deutsche Bank's Q2 holdings showed NVIDIA as its top position: 91.95 million shares (~$18.4B, 5.34% of portfolio).
NVIDIA's Prior AI Infrastructure Investments
Key Numbers
| Item | Value | |---|---| | Partner institutions | 6 (Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, KKR) | | Capital to mobilize | >$500B (long-term, third-party, not a single fund) | | NVIDIA residual value support | ≤25%, per-project | | H100 one-year rent | $1.70 → $2.35/GPU-hour | | Cross-cloud on-demand median | $2.00 → $2.70/GPU-hour | | B200 premium | $5.30–7.05/GPU-hour | | NVIDIA stock (day after) | -2.86% to $217.55, ~$5.3T cap | | 5Y CDS | +5.3 bp to 77.215 bp | | Apollo / Blackstone | +3.59% / +3.3% |
Three Layers of Significance
1. Compute is Revenue: NVIDIA's valuation logic shifts from data-center REITs (rent + depreciation) to productive assets (output and demand elasticity). 2. Systematic entry of financial capital into AI infrastructure: six major firms entering simultaneously is a new order of scale beyond BlackRock's earlier AI Infrastructure Partnership. 3. The circular financing debate: the CDS move shows not everyone buys the "virtuous cycle" narrative. The structure resembles early-2000s vendor financing.
Four Signals to Watch Before Year-End
Limitations and Unknowns
Sources