NVIDIA Invests $90 Billion Over Past 16 Months, Expanding AI Ecosystem Across 145 Companies

NVIDIA has invested about $90 billion over the past 16 months in more than 145 AI companies, spanning AI model development, cloud services, and infrastructure suppliers. According to Reuters data and PitchBook statistics, this spending accounted for 40% of NVIDIA's operating cash flow in the previous fiscal year.

What Happened

The Financial Times reported that in the year through January 25, NVIDIA committed about $47 billion in investments and partnership agreements, then added another $43 billion over the following four months. The largest single deal included a $20 billion partnership with Groq involving technology licensing and talent recruitment. NVIDIA also locked in component supply with $95 billion in spending commitments, such as investing $2 billion each in Coherent and Lumentum and obtaining $3.2 billion in warrants from Corning.

Mechanism Breakdown

These deals are mostly led by the business development team, rather than relying solely on NVIDIA's internal venture capital arm NVentures. The agreements often tie financing to technology collaboration; NVIDIA encourages portfolio companies to adopt its open-source AI model Nemotron, seeking to replicate the CUDA ecosystem model. Lawyers noted that technology teams and business development staff conduct two conversations at the same time: one about technology and one about financing.

NVIDIA said in an official response that the strategy aims to "create a large AI ecosystem, accelerating and diversifying innovation"; the investments do not come with exclusivity conditions, and customers choose based on product performance.

Industry Impact

For supply chain companies, NVIDIA's funding supports capacity expansion plans, strengthening their leverage when capacity is tight. Suppliers such as Coherent, Lumentum, and Corning are also major suppliers to Apple, and NVIDIA is now becoming a major customer through investment.

For AI model developers, obtaining NVIDIA funding means easier access to compute resources, but they must accept recommendations to use the Nemotron model. Disclosed cases include a $2 billion investment and NVLink compatibility agreement with Marvell, and a five-year, $3.4 billion GPU lease plus a $2.1 billion investment with Iren.

For developers and enterprise users, ecosystem expansion may accelerate the pace of innovation, but it also increases dependence on a single hardware platform. Regulators have issued broad information requests to NVIDIA concerning investments, partnerships, and agreements with foundation model developers.

Comparison and Precedents

This level of spending is far above the 6% of cash flow Alphabet devoted during the same period. Similar arrangements allow NVIDIA to simultaneously act as customer, supplier, and potential shareholder, with accompanying increases in financial and regulatory risk.

Strategic Assessment

Based on existing agreement patterns, more hybrid investment and leasing arrangements with new cloud service providers and chip design companies are the most likely next development.

These investments have already shifted NVIDIA in the AI era from a pure hardware supplier to an ecosystem shareholder and infrastructure provider. Founders of portfolio companies are gradually realizing that building products within NVIDIA's ecosystem can secure funding support, but it also brings potential antitrust scrutiny.