The race to build and finance AI infrastructure is reaching unprecedented levels, with two major developments underscoring the trend. Applied Compute, a California-based company specializing in open-weight AI model infrastructure, is reportedly in talks to secure funding at a valuation of approximately $3 billion, more than doubling its $1.3 billion valuation from a Series B round just four months ago. Meanwhile, Nvidia has announced a sweeping partnership with Wall Street heavyweights Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to mobilize over $500 billion in third-party capital for AI infrastructure projects.
Applied Compute, which has raised only $160 million to date, is attracting investor attention for its Agent Cloud (AC2) platform that enables enterprises to train, deploy, and continuously improve open-weight AI models on their own data and workflows. The company’s valuation leap reflects a broader shift toward customizable AI systems and the infrastructure that supports them, as evidenced by Hugging Face’s growth to 13 million users and over 2 million public models. Applied Compute’s research demonstrates cost efficiencies—its model router achieved GPT-5.5-level performance at roughly 25% lower cost. The funding discussions signal that investors see AI compute not merely as a utility but as an asset class, with the company’s $3 billion valuation based on relatively modest funding compared to giants like Mistral AI’s $13.7 billion.
The Nvidia-led initiative takes this financialization even further. By creating independent financing platforms, Nvidia aims to turn compute capacity, GPUs, and full-stack AI infrastructure into investable assets that can attract pension funds, insurers, and sovereign wealth funds. This move addresses the enormous upfront capital requirements of AI build-outs—Big Tech alone is expected to spend over $730 billion on AI-related capex this year, and Morgan Stanley estimates total AI infrastructure needs could reach $3.5 trillion between 2026 and 2028. The partnerships also tether the financed projects to Nvidia’s hardware and software ecosystem, raising questions about circular financing but offering a pipeline of long-term demand for AI compute.
Together, these developments point to a maturing AI infrastructure market where compute is being repackaged as a scalable, revenue-linked investment product, much like traditional infrastructure assets. The influx of institutional capital could accelerate the deployment of open models and enterprise AI, while also concentrating financial power among the companies and investors shaping this new asset class.
