Fluidstack as TPU Distribution Partner
Fluidstack
This potential Google check matters less as funding and more as channel control. Fluidstack is becoming one of the outside pipes through which Google can place TPU capacity into third-party data centers, alongside the company’s existing role as a builder and operator of very large GPU clusters. That turns Fluidstack from a cloud reseller into a strategic infrastructure partner that can help Google reach customers and geographies its own data center footprint does not cover.
-
Fluidstack already sits inside Google-backed infrastructure deals at large scale. It has hosting agreements with TeraWulf, Cipher, and Hut 8, with Google backstopping about $5.8B of lease obligations across those projects. That shows the TPU discussions fit a broader pattern, not a one off financing event.
-
The product logic is simple. Fluidstack runs clusters in third-party facilities and sells compute through its own console and contracts. If Google adds TPUs into that footprint, it gets another route to distribute custom chips beyond Google owned sites, while Fluidstack reduces its dependence on Nvidia hardware.
-
The valuation reflects this shift from broker to utility. Fluidstack was at an estimated $180M ARR in December 2024, with private cloud already 62% of revenue and much larger contract sizes than its marketplace business. Investors are paying for long duration infrastructure contracts and strategic chip distribution, not just spot GPU rentals.
From here, the company is heading toward a hybrid role as both neocloud operator and hardware route to market for hyperscaler chips. If that model works, AI infrastructure will split more clearly between chip makers, capital providers, and operators like Fluidstack that assemble power, buildings, hardware, and customer contracts into one deployable stack.