Featherless.ai raises 20 million dollars to scale open source AI platform
Featherless.ai has secured 20 million dollars in a Series A funding round to expand its open source artificial intelligence infrastructure. The company, based in San Francisco, develops a serverless inference platform designed to simplify and accelerate access to large scale AI models for developers and enterprises.
The funding round was co led by AMD Ventures and Airbus Ventures, with participation from BMW i Ventures and several existing investors. The deal reflects growing institutional interest in infrastructure layers that support deployment and execution of AI models rather than only model development. It also follows an earlier 5 million dollar seed round that helped the company establish its initial platform and expand its model catalog.
Featherless.ai has grown rapidly since its launch in late 2023. The company increased its library from around 4,000 open weight AI models to more than 30,000 across language, vision, and audio domains. Its core technical feature is a proprietary hot swapping system that allows models to be switched on GPUs in under five seconds, reducing downtime and improving computational efficiency for users running multiple workloads.
The platform processes around 6 billion tokens per day and has become a major provider of large language model inference through Hugging Face integration. It serves thousands of models directly through its infrastructure. In early 2026, the company introduced a managed runtime environment for AI agents, combining sandboxed execution with subscription based inference access, targeting developers building autonomous AI systems.
The new capital will be used to expand infrastructure capacity and support a shift toward fixed price inference pricing across the platform. The company aims to move away from traditional token based billing models used across much of the AI industry. Instead, it offers subscription plans starting at 10 dollars per month with usage defined by concurrency and model size rather than consumption volume, positioning itself as an alternative to major cloud AI providers.
-
13:26
-
13:21
-
13:06
-
13:06
-
13:02
-
11:30
-
11:23
-
11:20
-
11:20
-
11:13
-
11:11
-
11:06
-
11:05
-
10:59
-
10:59
-
10:56
-
10:54
-
10:41
-
10:38
-
10:35
-
10:31
-
10:22
-
10:05
-
09:51
-
09:33
-
09:06
-
09:00
-
08:56
-
08:50
-
08:43
-
08:37
-
08:30
-
08:25
-
08:14
-
08:02
-
07:51
-
07:43
-
07:36
-
07:30
-
07:28
-
17:25
-
17:23
-
17:21
-
17:18
-
17:16
-
17:14
-
17:11
-
17:07
-
17:01
-
16:55
-
16:50
-
16:23
-
16:02
-
15:53
-
15:47
-
15:46
-
15:16
-
15:01
-
14:57
-
14:52
-
14:49
-
14:37
-
14:19
-
14:00
-
13:53
-
13:44
