Fluidstack Eyes $1B Round at $18B Valuation After Sealing $50B Anthropic Data Center Deal

Fluidstack, the British AI neocloud startup, is in advanced talks to raise roughly $1 billion at an $18 billion valuation — more than double the $7.5 billion it commanded just four months ago. Jane Street and Situational Awareness are in discussions to co-lead the round, with Morgan Stanley advising, according to sources familiar with the negotiations.

The catalyst for the valuation leap is a $50 billion data center construction deal signed with Anthropic, covering facilities in Texas and New York. That single contract turned a mid-tier GPU cloud into one of the most consequential infrastructure plays in AI.

Fluidstack operates differently from hyperscalers. Instead of owning its own hardware, it aggregates compute capacity from third-party GPU owners and resells it to AI labs, startups, and enterprises. The model has proved lucrative as demand for inference and training capacity continues to outpace supply.

The new round, if closed, would bring Fluidstack’s total raised capital well past the $700 million it secured in its previous financing. The speed of the valuation climb — from $7.5B in December 2025 to a targeted $18B in April 2026 — reflects brutal competition for reliable AI compute that every model developer now experiences.

AI infrastructure is no longer a supporting act. Anthropic’s commitment at the $50 billion scale signals that even the most compute-hungry labs are diversifying beyond AWS, Google Cloud, and Azure. For Fluidstack, becoming a primary build partner for a top-three AI lab is a fundamentally different kind of moat than simply offering cheaper GPUs.

Jane Street’s presence as a potential co-lead is notable. The quant trading firm has been quietly deepening its exposure to AI infrastructure as a long-duration asset class, treating data center capacity as something closer to energy infrastructure than venture risk.

What developers should watch: if Fluidstack closes at $18B, it becomes a credible alternative to hyperscaler GPU reservations for teams building production inference workloads at scale. The firm’s pitch — stable capacity without long-term GPU contract lock-in — has become increasingly attractive as model context windows grow and token costs rise.

The deal isn’t closed yet. But a $50B build commitment from Anthropic and an $18B valuation signal that the GPU middleman business is graduating into something much harder to replicate.