Gimlet Labs, a startup whose software splits artificial intelligence workloads across different types of chips, has raised $300 million in a round that values the company at $3 billion, Bloomberg reported late Wednesday, citing an interview with Chief Executive Officer Zain Asgar.
Andreessen Horowitz led the fundraising. New backers include Arm Holdings and M12, Microsoft’s venture fund, Asgar told Bloomberg. Tech Funding News carried the same $300 million and $3 billion figures on Thursday and noted Gimlet’s prior $12 million seed and $80 million Series A.
Gimlet calls its product a multi-silicon inference cloud. The platform breaks inference jobs into pieces and routes each piece to the hardware best suited for it, including GPUs, CPUs, and specialized accelerators, rather than locking a workload to a single chip vendor. Asgar also told Bloomberg the company is collaborating with Arm to make its software compatible with various forms of Arm chip technology.
Tech Funding News said Gimlet is pushing beyond pure software into helping customers configure heterogeneous data centers, because different chips can need different cooling and rack layouts. The company reports three-to-ten-times inference speed improvements for the same cost and power envelope, according to that brief. Revenue and customer counts were not disclosed in the Bloomberg report.
The raise lands as investors pour money into AI infrastructure beyond a single GPU stack. This is a funding story about inference orchestration, not a model launch.