Singapore Co-Founded AI Startup Featherless.ai Secures $5M Seed Funding to Revolutionize Serverless AI Inference

Featherless.ai, a serverless AI inference platform co-founded by Singaporean entrepreneur Eugene Cheah, has raised $5 million in seed funding.

The funding round, announced in March 2025, was backed by prominent investors including Airbus Ventures, 500 Global, and Kickstart Ventures.

Democratizing Access to Open-Source AI

Featherless.ai aims to make running advanced open-source AI models faster and cheaper by tackling high compute costs in inference.

Inference, the process of using trained AI models for predictions, often bottlenecks developers due to expensive cloud infrastructure dominated by tech giants.

The platform's serverless model allows users to deploy, test, and fine-tune models at scale without managing servers, appealing to startups and researchers.

Singapore's Rising AI Talent on Global Stage

While based in San Francisco, Featherless.ai's Singapore roots highlight the Lion City's growing role in nurturing AI innovators who build globally competitive companies.

Eugene Cheah's background in efficient AI architectures like RWKV positions the startup to challenge inefficient transformer models prevalent today.

This funding comes amid booming demand for cost-effective AI tools, as open-source models like Llama proliferate but remain hard to scale affordably.

Industry Implications and Future Outlook

Investors like Airbus Ventures, typically focused on aerospace, signal AI inference's crossover potential in high-compute industries like aviation and manufacturing.

With the fresh capital, Featherless.ai plans to accelerate research into decentralized AI, reducing reliance on centralized big tech providers.

For everyday users, cheaper inference means more innovative apps—from personalized tutors to efficient logistics—without prohibitive costs.

As AI adoption surges, startups like Featherless.ai could level the playing field, fostering broader economic growth in Southeast Asia and beyond.