Gimlet Labs, an AI inference infrastructure startup, has raised $300 million in a Series B round, securing a $3 billion valuation. The company's multi-silicon cloud technology distributes AI workloads across chips from six manufacturers, aiming to optimize performance and cost.
TL;DR
- Gimlet Labs raises $300M at $3B valuation, led by Andreessen Horowitz, with participation from Arm, Microsoft's M12, and Samsung Ventures
- The company's multi-silicon cloud technology supports chips from Nvidia, AMD, Intel, Arm, Cerebras, and d-Matrix, aiming to improve AI inference efficiency
- Gimlet Labs has expanded into data center infrastructure, securing multi-billion-dollar contracts and scaling managed capacity to several hundred megawatts
What happened
Gimlet Labs, founded in 2023 and headquartered in San Francisco, has raised $300 million in a Series B round, bringing its total funding to $392 million. The round was led by Andreessen Horowitz, with participation from Arm Holdings, Microsoft's M12 venture arm, and Samsung Ventures. The company has achieved a $3 billion valuation, despite not designing its own AI chips, focusing instead on software and infrastructure technology.
Gimlet Labs' core offering is its multi-silicon inference cloud, which decomposes AI workloads into multiple stages and allocates each to the most suitable hardware. The platform currently supports chips from six semiconductor manufacturers: Nvidia, AMD, Intel, Arm, Cerebras, and d-Matrix. The company claims that distributing inference workloads across heterogeneous silicon can boost performance by 3–10x under identical cost and power conditions.
In addition to its inference software, Gimlet Labs has expanded into data center infrastructure. The company is supporting customers' data center buildouts and expanding its own infrastructure, securing multi-billion-dollar contracts and scaling managed capacity to several hundred megawatts since March. Gimlet Labs is also collaborating with Arm to run its software on Arm-based semiconductors.
Why it matters
Gimlet Labs' multi-silicon approach addresses the varying hardware characteristics required for different AI inference stages, potentially reducing cost and power waste inherent in single-chip architectures. This strategy is particularly relevant as AI agents exhibit unpredictable workload patterns, alternating between short responses and long reasoning chains.
The company's expansion into data center infrastructure is a response to the unique power densities, networking, and cooling requirements of operating multiple chip types in a single facility. By addressing these challenges, Gimlet Labs aims to support the growing demand for AI inference and data center power constraints.
However, structural risks exist, as partners like Nvidia, AMD, Intel, and Cerebras could build their own chip orchestration layers, potentially undermining Gimlet Labs' core promise of running anywhere. Additionally, the company has not disclosed customer count or revenue, making its $3 billion valuation largely a bet on future demand.
Key facts
- Gimlet Labs raised $300 million in a Series B round, securing a $3 billion valuation, according to finance.biggo.com
- The round was led by Andreessen Horowitz, with participation from Arm Holdings, Microsoft's M12 venture arm, and Samsung Ventures
- Cumulative funding for Gimlet Labs now totals $392 million, per the company
- The company's multi-silicon inference cloud supports chips from Nvidia, AMD, Intel, Arm, Cerebras, and d-Matrix
- Gimlet Labs claims its technology can boost AI inference performance by 3–10x under identical cost and power conditions
- The company has expanded into data center infrastructure, securing multi-billion-dollar contracts and scaling managed capacity to several hundred megawatts since March
- Gimlet Labs is collaborating with Arm to run its software on Arm-based semiconductors
- The company was founded in 2023 and is headquartered in San Francisco, originating from a Stanford University research project
Context
Gimlet Labs' raise is part of a broader wave of large-scale AI infrastructure financing this year. Last month, a company that pivoted from Bitcoin mining to AI cloud infrastructure raised $3 billion at a $30 billion valuation, reflecting the growing investment in AI infrastructure.
The AI inference market is expected to grow significantly in the coming years, driven by the increasing demand for AI applications and the need for efficient and cost-effective inference solutions. Gimlet Labs' multi-silicon approach addresses these market needs by optimizing performance and cost across heterogeneous hardware.
As AI agents become more prevalent, their unpredictable workload patterns present a challenge for traditional single-chip architectures. Gimlet Labs' product thesis of sitting atop all chips rather than betting on a single architecture is a key logic underpinning its valuation and market potential.
