FluidStack is in talks to raise $1B at an $18B valuation

At a glance:

• FluidStack negotiating $1B raise at $18B valuation, up from $7.5B in December.
• Jane Street and Situational Awareness co-leading; Morgan Stanley advising.
• $50B Anthropic data centre deal in November 2025 drove valuation surge.

Record-breaking valuation talks signal AI infrastructure boom

FluidStack, the Oxford-founded neocloud startup specialising in AI compute infrastructure, is in advanced discussions to raise approximately $1 billion at a target valuation of $18 billion, according to Bloomberg. The talks, which involve trading firm Jane Street and Situational Awareness—an AGI-focused fund led by former OpenAI researcher Leopold Aschenbrenner—mark a dramatic acceleration in the company's valuation trajectory. Morgan Stanley is advising on the investment, underscoring the scale and complexity of the transaction.

The $18 billion target represents more than a doubling of FluidStack's worth in just a few months. In December 2025, Bloomberg reported that the company was negotiating a $700 million raise at a $7.5 billion valuation, with Situational Awareness expected to lead that round. That earlier deal never formally closed or received public confirmation, but the new talks suggest FluidStack's negotiating position has strengthened considerably since then.

Anthropic partnership fuels exponential growth

The catalyst for this valuation surge is almost certainly FluidStack's $50 billion data centre partnership with Anthropic, signed in November 2025. Under the agreement, FluidStack will build custom facilities in Texas and New York specifically optimised for Anthropic's Claude AI workloads, with sites coming online throughout 2026. This represents the largest infrastructure commitment Anthropic has made independently from major cloud providers such as AWS and Google.

The partnership marked a defining moment for FluidStack, transforming it from a relatively unknown European AI infrastructure player into a major force in the global AI compute market. Following the deal's signing, the company relocated its headquarters from the UK to the United States, positioning itself closer to the epicentre of AI development and investment.

From Oxford spin-off to AI infrastructure powerhouse

FluidStack was founded in 2017 as a spin-off from Oxford University, initially pursuing different technological applications before pivoting aggressively toward AI compute infrastructure following the release of ChatGPT in late 2022. This strategic shift proved prescient as demand for specialised AI compute exploded.

The company's growth has been exponential. Revenue climbed from $1.8 million in 2022 to $66.2 million in 2024 as FluidStack built out capacity for AI customers. The startup now manages more than 100,000 GPUs and counts major AI players among its customers, including Anthropic, Meta, Poolside, Black Forest Labs, and France's Mistral.

Strategic partnerships and massive infrastructure projects

A $200 million Series A in February 2025 accelerated FluidStack's expansion, providing capital for further infrastructure development. In early 2025, the company was also named the primary partner for a one-gigawatt AI infrastructure project backed by the French government, representing more than $11 billion in spending.

Additionally, FluidStack has secured two 10-year hosting agreements with TeraWulf totalling $6.7 billion in contracted revenue, set to begin in 2026. These agreements, combined with the Anthropic partnership, position FluidStack among the most valuable AI infrastructure companies outside the hyperscaler tier.

Why bespoke AI compute matters now

The valuation trajectory—from startup to $18 billion in under a decade—reflects the intensity of demand for bespoke AI compute at a moment when frontier AI labs are racing to secure infrastructure that hyperscalers cannot build fast enough or customise sufficiently for specific workloads. Unlike general-purpose cloud providers, FluidStack builds data centres optimised for the training and inference patterns of specific AI systems, giving customers more control over cost and performance than rack rental from AWS or Google Cloud provides.

This specialisation has become increasingly valuable as AI models grow more complex and computationally intensive. The ability to design facilities around specific AI workloads, rather than offering generic compute resources, has positioned FluidStack as an essential partner for companies pushing the boundaries of artificial intelligence.

What's next for FluidStack

If the current funding round closes at the reported terms, FluidStack would join an elite group of AI infrastructure companies commanding valuations in the tens of billions. The capital would likely fuel further expansion of its data centre footprint, both in the United States and internationally, as well as continued investment in optimisation technologies that differentiate its offering from traditional cloud providers.

The company's rapid ascent also highlights broader trends in the AI industry: the growing importance of specialised infrastructure, the willingness of investors to back infrastructure plays at massive valuations, and the strategic value of partnerships between infrastructure providers and AI model developers. As frontier AI labs continue their infrastructure arms race, companies like FluidStack that can deliver customised, high-performance compute solutions are likely to remain in high demand.