Why NVIDIA Says AI Compute Should Be Treated as an Investable Asset

Jensen Huang announced on X a collaboration with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to set up an independent compute‑financing platform that aims to mobilise more than $500 billion of third‑party capital for global AI infrastructure.

The partnership seeks to shift AI compute from the traditional model of enterprises buying GPUs and building data centres to a financing model similar to power, telecom and transport infrastructure, treating AI factories as investable assets that generate revenue.

Huang emphasized that the $500 billion figure is not NVIDIA’s revenue nor a single fund size; each financial institution will independently evaluate client demand, utilisation, cash flow and residual value, with NVIDIA potentially providing up to 25 % residual‑value support on a project‑by‑project basis.

A New Infrastructure Asset

NVIDIA’s AI factory comprises GPUs, networking, system software, AI frameworks and a global developer ecosystem. The DSX AI factory can run a wide range of models across language, vision, speech, biology, physics and robotics, serving multiple customers simultaneously.

Because the architecture is widely adopted by cloud providers and system vendors, an AI factory can be redeployed to different customers, preserving residual value. CUDA software upgrades continuously improve performance and efficiency, extending the economic life of the hardware.

Example: the A100 GPU, launched in 2020 on the Ampere architecture, remains in commercial AI training, fine‑tuning and inference workloads six years later, with multi‑year compute‑deployment contracts pushing its useful life toward ten years.

Market lease prices illustrate resilience: one‑year H100 lease rose from $1.70 per GPU‑hour (Oct 2025) to $2.35 (Mar 2026); on‑demand median price increased from $2.00 to $2.70 per GPU‑hour (Oct 2025–Jun 2026). Blackwell B200 cloud lease ranges $5.30‑$7.05 per GPU‑hour.

Capital Entering AI Factories

The demand for AI infrastructure is strong but uneven; many AI labs, startups and cloud providers lack the scale or capital to build AI factories. The six financial partners, experienced in infrastructure investment, will help eligible projects obtain larger‑scale financing.

Each partner will independently assess projects on demand, utilisation, cash flow and residual value; NVIDIA provides the platform while the financiers supply long‑term capital.

Key Questions

Is this circular financing? The design addresses concerns by seeking independent, long‑term institutional capital rather than internal industry financing loops.

Why does NVIDIA support financing? NVIDIA may offer up to 25 % residual‑value support per project, but only as a supplement to the financiers’ own risk assessments.

Can the market absorb the compute? The focus is on building AI factories that continuously create economic value, not merely adding more data centres.

Where does ROI come from? From the practical value of AI itself—software development, drug discovery, product design, customer service, automation—supported by the AI factory infrastructure.

Infrastructure for the Intelligent Age

Historically, major industrial revolutions relied on large‑scale infrastructure funded by external capital. NVIDIA views AI factories as the new foundational infrastructure for the intelligent era and aims to make them accessible through this financing mechanism.