"If you look at the AI landscape, the US and China collectively represent less than half of the world. I don't want a world economy where only half of its participants can use this technology. That is why we focus heavily on researching multilingual open-source AI—to make it accessible for the rest of the world and ensure nobody is left behind." - Eugene Cheah
"We heard the same 'winner-takes-all' narrative when IBM Db2 first launched as a closed-source database. Instead, the outcome was a fragmented landscape where enterprises used multiple databases, each with its own pros and cons. That exact same scenario will play out for AI. Once the open-source models become good enough, companies find it more reliable and sustainable to use them over expensive, closed-source versions." - Eugene Cheah
"Within Asia, the Philippines felt the impact of AI the most because they are the remote support center of the world. Their call centers were hit heavily as these were the first services automated by AI. My advice to existing industries is to start augmenting with AI now to scale your capabilities, retain your customers, and protect your GDP before those jobs disappear to fully automated systems." - Eugene Cheah
00:00 - Introduction
01:57 - From UI Testing to AI Breakthrough: How solving a UI testing problem led to discovering a highly efficient, open-source alternative to transformer models.
07:13 - Spinning Out Featherless AI: The decision to launch a new entity dedicated to AI inference and raising $1 million in San Francisco with no pitch deck.
16:15 - Open Source vs. Closed Source AI: Why the AI race will mirror the database industry, leading to a fragmented ecosystem rather than a "winner-takes-all" monopoly.
21:52 - China's Open Source Strategy: How chip restrictions forced Chinese companies into rapid innovation and an incidental open-source AI boom.
28:06 - The Value of Specialization: Why businesses need access to thousands of highly specialized, localized models instead of relying on constantly changing generalist APIs.
32:13 - Shifting the Benchmark: Why the future of enterprise AI isn't about frontier intelligence, but absolute reliability for specific, localized tasks.
39:01 - AI's Threat to Southeast Asian Economies: The immediate impact of AI automation on BPO hubs like the Philippines and service-based economies like Singapore.
43:39 - Preventing a Global AI Divide: The importance of multilingual AI in preventing a scenario where non-English and non-Chinese speaking populations are economically excluded.
Keywords: Open Source AI, AI Inference, Southeast Asia Tech, Multilingual AI Models, Singapore Startups, Philippines BPO Automation, AI Policy and Geopolitics
