SEA’s AI firms take the battle elsewhere

Sign up for the Daily Newsletter, sent exclusively to our premium subscribers. We break down the big and messy topics of Asia’s tech and startup community. Get the newsletter in your inbox everyday with a premium subscription.

Hello reader,

In 2012, a Counter-Strike team called Ninjas in Pyjamas went on an 87-0 winning streak. For all you football fans out there, this was the gaming equivalent of Arsenal’s unbeaten run from 2003 to 2004.

Teams like these have always impressed me. While boring in some ways, there is some fun in trying to guess when that streak will eventually break – no king rules forever.

In the AI world, OpenAI’s dominance in large language model (LLM) development represents a similar scenario. Since it stormed onto the scene with ChatGPT in late 2022, OpenAI has more or less been the undisputed leaders in the space.

Until now.

With DeepSeek shaking things up, it’s become clear that competing in LLM development may not be the path to success for Southeast Asian AI startups, as these giants may just be too far ahead. Instead, it might be better to build on top of the tech and target end users instead. Our story for today explores how the region’s startups can do so.

Today we look at:

• The next step for AI development in Southeast Asia
• China’s new fund for tech startups
• Other newsy highlights such as JD.com’s profitable Q4 and Tesla’s waning sales in mainland China

Premium summary

Rolling in the deep

Image credit: Timmy Loen

Everyone wants to be the best, but sometimes it’s good to recognize when others are too far ahead. In that case, it’s time to focus on where you can actually add value. For Southeast Asia’s AI startups, that means building on top of existing LLMs to offer better services and solutions for end users.

• User priorities: Many users won’t really care which LLMs are being used, so regional startups building their own foundational models won’t make sense. What users will focus on is the apps built on top of these platforms.
• Under the SEA: That said, it’s not like regional LLMs won’t matter. Developing them can help minimize biases in models and enhance understanding of underrepresented languages, among other benefits.
• Nothing to fear: Previously, one fear for AI startups was that foundational model providers would hike costs, putting a strain on margins. However, the introduction of DeepSeek has made such price controls more difficult for big players to implement.

Read more: SEA’s AI players target end users as DeepSeek stifles LLMs

Throwing money at the problem

Start seeing real results at #TIAConference2025

Quick bytes