*How This $3.6M/yr Startup Runs Open AI Models* Featherless.ai gives developers instant access to over 6,700 open source AI models on a flat rate subscription, no per-token billing, no GPU setup, no engineering overhead. Over 10,000 customers today. Largest customer pays $1 to $2 million per year. Eugene started this as a pricing experiment inside a different company. Within a few days it was generating more revenue than the original product, so they became Featherless. Today they're doing over $250,000 a month and scaling toward $500,000, with a $20 million Series A closed in December 2025 and a team of 27 people running infrastructure for some of the most compute-intensive AI workloads in the world. --- *In this episode:* - How Featherless went from a pricing experiment to $3M+ in annual revenue in under two years - Why their largest contracts are $1 to $2 million per year even though their entry plan is $25 a month - How they cut AI inference costs by up to 10x for startups burning $100,000 a month on OpenAI or Anthropic - Why they support 6,700 models when every other provider stops at 100, and why that is the moat - How they grew to 10,000 customers with zero paid marketing using only Reddit and word of mouth --- š *Get funding at Founderpath:* https://founderpath.com/ --- *Featherless Key Metrics (2026)* *Revenue:* $250,000 to $500,000 per month (recorded April 2026) *Customers:* 10,000+, largest paying $1 to $2M per year *Product Scope:* Serverless inference for 6,700+ open source AI models on Hugging Face *Pricing Model:* $25/month entry plan, scales to enterprise contracts in the millions annually *Funding:* $2M seed (pre-2023) + $20M Series A (December 2025) --- *How Featherless is Growing* - Reddit and Hugging Face word of mouth drive all inbound, zero paid marketing as of April 2026 - Enterprise sales motion targets startups spending $100K+ per month on OpenAI or Anthropic and cuts their costs in half or more - Model breadth is the moat, supporting 6,700 models versus competitors' 100 makes them the only viable option for companies fine-tuning their own AI - Self-upgrading infrastructure: 12 of 27 employees work full-time on infra, keeping costs low enough to pass savings directly to customers --- *Subscribe for more SaaS revenue breakdowns:* *Spotify:* https://open.spotify.com/show/3AA9buFn8oxJf8WSDOUks5 *Apple Podcasts:* https://podcasts.apple.com/us/podcast/saas-interviews-with-ceos-startups-founders/id1023761733 *Visit:* https://getlatka.com/companies/tive.com --- *Timestamps:* 00:00 $3.6M revenue from open source AI 00:47 What Featherless actually does 01:52 Explain it like I'm five 03:54 10,000 customers at $25 a month 05:42 Started as a pricing experiment 07:18 Zero to $250K a month in months 09:28 Cutting OpenAI bills in half 13:00 Why they support 6,700 models 16:05 Largest customer pays $2M a year 17:18 The future of AI architecture --- #saas #bootstrapped #revenue #featherless