Fluidstack Shifts to US AI Infrastructure

Moving the headquarters to New York means Fluidstack is no longer selling mainly as a fast European GPU broker, it is positioning itself as a builder of large, long term American AI infrastructure. That fits the company’s shift from short term GPU rentals toward Private Cloud contracts, where customers reserve dedicated clusters for years and often pay a meaningful portion up front. The US is where the biggest AI labs, financing pools, and hyperscale data center builds are now concentrated.

• Fluidstack’s business mix already points this way. By the end of 2024, about 62% of revenue came from Private Cloud, versus 38% from its original marketplace. Private Cloud carries much higher gross margins, 85% versus 13%, and uses multi year contracts tied to dedicated hardware.
  2 sacra 3 sacra
• The Anthropic partnership turns the move into operating reality, not just symbolism. Anthropic announced a $50B US compute buildout with Fluidstack facilities in Texas and New York, with sites expected through 2026. That gives Fluidstack an anchor tenant for domestic capacity and a path into more reserved capacity deals with frontier model labs.
  1 sacra 4 sacra
• This also changes who Fluidstack competes with. Earlier research placed it closer to Lambda and Together AI at the startup end of GPU cloud. A New York headquarters and US data center program push it nearer to CoreWeave and Crusoe, companies that win by financing and operating large dedicated clusters for the biggest model builders.
  2 sacra 3 sacra

The next step is a full transition from regional GPU marketplace to transatlantic compute utility. If Fluidstack keeps landing US lab and enterprise contracts, New York becomes the commercial hub for bigger reserved capacity deals, while its European footprint remains a differentiator for sovereign AI projects. That combination could make it one of the few neoclouds with real footing on both sides of the Atlantic.