Fluidstack, a startup constructing specialized data centers for AI workloads, is negotiating a $1 billion funding round that would value the company at $18 billion, according to a Bloomberg report. This potential deal, possibly led by trading firm Jane Street, represents a dramatic valuation jump from the $7.5 billion figure reported just months prior in December.

The surge in investor interest follows a pivotal technical partnership. In November, Anthropic committed $50 billion to have Fluidstack build custom data centers in Texas and New York. This move signals a strategic shift for AI labs: while Anthropic relies on AWS and Google Cloud for inference, its explosive growth demands dedicated, purpose-built capacity for training. Fluidstack's architecture diverges from general-purpose hyperscalers by optimizing its entire stack—power, cooling, and networking—specifically for dense AI compute clusters.

For engineering teams, the model presents a clear trade-off. It offers deeper hardware control and potential performance gains over shared cloud instances but requires managing a more complex, multi-vendor infrastructure footprint. Fluidstack has already reorganized its operations around this bet, moving its headquarters from the UK to New York and reportedly exiting a major European project to concentrate on U.S. builds. The company's customer list, which includes Meta and several other AI firms, suggests a growing market for infrastructure that prioritizes bespoke design over cloud generality.

Source: TechCrunch