Featherless.ai raises $20M Series A for serverless open-source inference platform

Featherless.ai raises $20M Series A for serverless open-source inference platform

Featherless.ai, a San Francisco-based AI infrastructure company providing a serverless inference platform for open-source models, has raised $20 million in a Series A round co-led by AMD Ventures and Airbus Ventures. The round also included BMW i Ventures, Kickstart Ventures, Panache Ventures, and Wavemaker Ventures. The company plans to use the capital to scale its global infrastructure, launch a marketplace for specialized open models, and deepen integration with diverse hardware architectures.

Why it matters: This raise signals the continued arms race in the inference layer for open-weight models, a segment that is increasingly contested as foundation labs release more performant open-source models. Featherless.ai positions itself as a hardware-agnostic, serverless alternative to hyperscaler-proprietary inference services. AMD Ventures' involvement underscores the chipmaker's strategy to build an ecosystem around its GPUs, challenging Nvidia's dominance in AI inference. However, the $20M sum is modest relative to the capital needed to break into the crowded AI infrastructure landscape, where CoreWeave and others have raised billions.

Grounded expert take: The move is consistent with the broader pattern of open-source model distribution and hardware diversification. Featherless.ai's support for over 30,000 models and its RWKV architecture pedigree give it technical credibility, but the real challenge will be winning enterprise trust and compute scale against cloud giants. The participation of AMD Ventures is a telling cross-substrate signal — AMD is investing to seed its GPU ecosystem in inference workloads. Still, without a massive compute allocation, Featherless.ai may remain a niche player for developers seeking flexibility.

#FeatherlessAI #AIInfrastructure #ServerlessInference #OpenSourceAI #AMDVentures #SeriesA