FriendliAI is a technology company.
FriendliAI has raised $20.0M across 1 funding round.
FriendliAI has raised $20.0M in total across 1 funding round.
FriendliAI has raised $20.0M in total across 1 funding round.
FriendliAI's investors include 7percent Ventures, AI Fund, Alumni Ventures, Hoxton Ventures, Mosaic Ventures, Pi Labs, Point Nine Capital, RTP Global, Sierra Ventures, Two Sigma Ventures, Brian Sugar, Esteban Ordano.
FriendliAI is a generative AI infrastructure company specializing in high-performance inference platforms for large language models (LLMs) and custom AI agents. Its flagship product, Friendli Engine, delivers the fastest inference on the market through optimizations like continuous batching, custom GPU kernels, FP8/INT8 quantization, Multi-LoRA serving, and speculative decoding, enabling 2×+ faster speeds, ultra-low latency, 99.99% uptime SLAs, and up to 50% GPU cost reductions.[1][4][5] The platform serves enterprises and AI startups—such as Twelve Labs, Upstage, SKT, DeepAI, NextdayAI, Wrtn, ScatterLab, and Tunib—solving production challenges in deploying scalable, cost-effective generative AI for applications like chatbots, document analysis, and real-time agents.[2][4] With $6.74M raised since 2021 and integrations like Hugging Face (supporting 490,000+ models), FriendliAI shows strong growth momentum, including partnerships and enterprise deployments across the US and APAC.[3][4][5]
FriendliAI was founded in 2021 by Byung-gon Chun, a prominent AI researcher and inventor of *Continuous Batching* (also called Iteration Batching), a breakthrough technique for improving GPU utilization in LLM serving.[1][3][5][8] Based initially in Seoul, South Korea, with current operations in Redwood City, California, the company emerged from Chun's expertise in optimizing generative AI inference amid rising demands for efficient LLM deployment.[3][5][8] Early traction came from its developer-friendly serverless endpoints and Hugging Face integration, enabling seamless scaling from pay-as-you-go to enterprise containers; pivotal moments include securing investments from Sierra Ventures and building a customer base in Korea and the US, amplified by hires like Sang Won Lee in 2025.[3][5]
FriendliAI rides the explosive growth of generative AI inference, a market projected to exceed $250B by 2030, driven by surging demand for LLMs in agents, chatbots, and enterprise apps amid abundant but underutilized GPUs.[4][5] Its timing is ideal post-ChatGPT, when inference bottlenecks—high latency, costs, and unreliability—emerged as key hurdles for scaling AI from prototypes to production.[1][2] Market forces like Hugging Face's open ecosystem (390,000+ models) and multi-cloud GPU availability favor FriendliAI's optimizations, positioning it against competitors like Fireworks AI and Chalk by enabling APAC-US bridge and developer adoption.[3][5] It influences the ecosystem by powering reliable AI agents (e.g., via MCP for structured outputs) and fostering custom model innovation, reducing barriers for startups and enterprises.[4][6][7]
FriendliAI is poised to dominate AI inference as agentic AI proliferates, with expansions into complex multi-agent systems, deeper AWS/Hugging Face ties, and global enterprise wins leveraging its APAC roots and US momentum.[4][5][7] Trends like rising token volumes to trillions, edge inference, and cost pressures will amplify its edge in efficiency and reliability, potentially driving acquisitions or massive funding rounds. Its influence could evolve from infrastructure enabler to standard-setter for production AI, unlocking generative AI's full potential for businesses—echoing its founding mission to make elite inference accessible and scalable.[1][5]
FriendliAI has raised $20.0M across 1 funding round. Most recently, it raised $20.0M Seed in August 2025.
| Date | Round | Lead Investors | Other Investors |
|---|---|---|---|
| Aug 1, 2025 | $20.0M Seed | 7percent Ventures, AI Fund, Alumni Ventures, Hoxton Ventures, Mosaic Ventures, Pi Labs, Point Nine Capital, RTP Global, Sierra Ventures, Two Sigma Ventures, Brian Sugar, Esteban Ordano, Gokul Rajaram, Matthew Hackett, Omid Ashtari, Paul Forster, Pierre Lavaux, Trevor McFedries |