Is NVIDIA 'buying' the entire AI ecosystem? Over $40 billion invested in just a few months
NVIDIA is not only the "queen" of semiconductors in the field of artificial intelligence—i.e., the company that is capitalizing the most on this new technological phenomenon—but is increasingly becoming a central investor in the entire supply chain connected to it.
According to reports from CNBC, the company has already committed over $40 billion in equity investments in the first months of 2026, a figure that significantly surpasses the pace recorded last year. The most substantial part of this total comes from the $30 billion investment in OpenAI, NVIDIA's long-time partner and developer of ChatGPT. This move reaffirms the strong relationship between the two companies, which has been further strengthened since the explosion of the generative AI market in 2022.
In parallel, NVIDIA has intensified its stakes in publicly traded companies active in key sectors of AI infrastructure. In recent days, the company announced an agreement with IREN that could lead to an investment of up to $2.1 billion, while it has signed a deal with Corning that allows for a potential investment of up to $3.2 billion.
In the case of IREN, the agreement includes the distribution of NVIDIA's DSX infrastructures to power AI workloads of up to 5 gigawatts in globally distributed data centers. Corning, on the other hand, will build three new production facilities in the United States dedicated to optical technologies for NVIDIA, in a context where the industry is progressively leaning towards fiber-optic connections for next-generation rack-scale systems.
The group’s investments also extend to photonics. In March, NVIDIA invested $2 billion in Marvell Technology to collaborate on the development of silicon photonics-related technologies. During the same period, similar investments were announced in Lumentum Holdings and Coherent Corp., both active in the development of optical components for high-speed AI networks.
Another strategic avenue concerns the so-called "neocloud", or cloud operators specialized in artificial intelligence. NVIDIA has invested $2 billion in CoreWeave to expand data centers based on its technologies, alongside allocating another $2 billion to Nebius Group in a deal related to AI infrastructures, fleet management, and designing AI factories.
However, this strategy raises questions among analysts and investors. Some observers believe that NVIDIA is directly financing companies that, in turn, purchase its GPUs and systems, fueling a model of "circular investments." The comparison made by some experts is with the vendor financing practices that contributed to the dot-com bubble in the early 2000s.
Matthew Bryson of Wedbush Securities noted that NVIDIA's investments fall "fully within the theme of circular investments," while acknowledging that such moves could enhance the company's competitive advantage if the industrial execution proves effective. More critical is Jordan Klein of Mizuho, who warns that some agreements with cloud operators could risk appearing as pre-financing for the purchase of NVIDIA GPUs themselves.
For his part, CEO Jensen Huang defended the company’s approach, explaining that NVIDIA aims to support the entire AI ecosystem without "picking winners," investing in multiple companies active in the development of foundation models and infrastructures.
The numbers do show significant growth in the group’s financial exposure. In the last fiscal year, NVIDIA invested $17.5 billion in private companies and infrastructure funds, while the value of its unlisted holdings on the balance sheet increased from $3.39 billion to $22.25 billion. The company also recorded strong capital gains on its equity holdings, including its stake in Intel, whose stock saw over 200% growth in the last year.
NVIDIA's next quarterly report will likely provide a more detailed picture of the economic impact of this expansion strategy, which clearly aims to consolidate control not only over the GPU market but over the entire global infrastructure of artificial intelligence.