Crusoe and Fluidstack Raise $4.5 Billion to Lead AI Infrastructure Funding Wave

The commercial viability of artificial intelligence currently rests entirely on a physical constraint: the availability of specialized data centers equipped with enough power and cooling to run thousands of advanced graphics processing units simultaneously. That physical bottleneck is now pulling venture capital away from software applications and into heavy construction, evidenced this week by $4.5 billion in new funding for two companies that build and operate the facilities where AI actually lives.[1][2]

Denver-based Crusoe led the week's capital influx, closing a $3 billion Series F funding round that tripled its valuation to $30 billion in less than a year. New York-based Fluidstack followed immediately with a $1.5 billion private equity round, doubling its own valuation to $18 billion since its previous raise in July 2026. Together, the two transactions represent one of the largest single-week capital deployments into the compute layer—the physical infrastructure sitting between the silicon chips and the software models—in the industry's history.[1][2][5]

The driving force behind both valuations is quantitative trading firm Jane Street Group, which has emerged as a dominant financier of the AI infrastructure buildout. Jane Street led Fluidstack's $1.5 billion round and simultaneously signed a five-year, $13 billion cloud computing contract with Crusoe. Under the agreement, Crusoe will provide the trading firm with dedicated clusters of GPUs and the supporting infrastructure required for AI training and inference.[4][5]

The $13 billion commitment makes Jane Street the highest-profile customer for Crusoe's cloud business to date. It also signals that proprietary trading desks are now competing directly with major technology laboratories for raw computing capacity. The Crusoe contract reportedly served as collateral to help attract the $3 billion in equity funding, demonstrating how guaranteed, long-term cash flows from financial institutions are underwriting the massive capital expenditures required to build new data centers. Because the transactions were conducted privately, no executives from Crusoe, Fluidstack, or Jane Street are quoted in the financial filings, and representatives for all three firms declined to comment publicly on the valuations.[3][4]

Fluidstack's trajectory illustrates the sector's rapid evolution from software aggregation to physical asset management. Founded in 2017 as an Oxford University spinout that stitched together idle GPUs for researchers, the company has pivoted entirely to developing and operating purpose-built AI data centers. The startup is now the first publicly known operator of Google's proprietary Tensor Processing Unit capacity outside of Google itself, helping the search giant deploy its silicon to enterprise customers.[1][3]

That strategic position secured Fluidstack a central role in a $50 billion American compute program announced by Anthropic in November 2025. Backed by Google, the initiative tasked Fluidstack with building custom data center sites in Texas and New York to house the infrastructure needed to train the next generation of Anthropic's Claude models. The sheer scale of these deployments has forced Fluidstack to recruit heavily from aerospace and automotive manufacturers, shifting its hiring bar from cloud software engineering to heavy construction and industrial power systems.[1][3]

Crusoe underwent a similar industrial transformation. The company originally built its business by capturing excess, flared natural gas from energy fields to power cryptocurrency mining operations. It has since repurposed that distributed power expertise to develop massive AI data center projects, including the flagship Texas site for Oracle and OpenAI's Stargate project. In March 2026, Microsoft agreed to lease an expansion of that same Stargate facility, cementing Crusoe's transition from a niche energy startup to a primary infrastructure provider for the world's largest technology companies.[2][4]

The speed at which Crusoe and Fluidstack have been repriced reflects a structural shift in how capital markets view AI. Investors are increasingly treating GPU cloud providers, known as neoclouds, as long-duration infrastructure assets rather than traditional venture capital bets. As enterprise demand for computing power continues to outstrip the capacity of legacy cloud providers, the companies capable of securing the land, power, and chips are capturing the vast majority of the sector's new funding.[1][2][4][5]