- Premium Content It takes our newsroom weeks - if not months - to investigate and produce stories for our premium content. You can’t find them anywhere else.
Glenn Kaonang · · 6 min readSEA’s AI players target end users as DeepSeek stifles LLMs
The meteoric rise of Chinese startup DeepSeek underscores a harsh reality in AI: Competition is so intense that even well-funded giants like OpenAI struggle to maintain an edge.
US-based OpenAI had been offering a model with advanced reasoning – considered the next big step in AI development – through its US$200-a-month ChatGPT Pro plan until DeepSeek released its reasoning model for free in January.
In response, OpenAI integrated a similar feature into its free version of ChatGPT. In a short span, DeepSeek has ramped up competition in the AI landscape, pushing both Western and Chinese tech bigwigs to accelerate their advancements.
Image credit: Timmy Loen
With the AI’s industry leading giant feeling the heat, what does this mean for smaller players in Southeast Asia? DeepSeek’s entry might have raised the stakes, making it harder for the region’s homegrown large language models (LLMs) to compete.
Industry players believe that for many regional AI startups, the smarter move might be to build on top of existing LLMs. They could possibly even go the route of agentic AI – systems that can act independently to perform complex tasks – rather than competing at the foundational level.
“All the necessary frameworks [for agentic AI] are already available. The real game-changer is finding the right use cases to focus on,” says Irzan Raditya, co-founder and CEO of Kata.ai, a B2B AI firm based in Indonesia.
Can’t compete
As AI heavyweights continue to raise the bar, LLMs are becoming less of a unique selling point and more of an underlying infrastructure that powers AI-driven applications. LLMs are heading in the same direction as cloud computing, where users care more about the applications built on top of platforms than the platforms themselves.
Kata.ai’s Raditya doesn’t see a compelling reason for most AI startups in Southeast Asia to focus on building foundational models.
Users prioritize functionality over who provides the model. For example, most users didn’t even notice when Indonesian super app GoTo migrated its services from Google Cloud to Alibaba Cloud.
See also: Chips stacked against SEA? New rules may hinder AI boom in region
Kata.ai was founded in 2015, when LLMs were still largely unknown and AI was more commonly associated with natural language processing (NLP). At the time, local efforts were relevant because NLP was primarily solving language-related challenges.
But advanced LLMs are widely available now, so language comprehension is no longer a major barrier – not even for the colloquial Bahasa Indonesian used in everyday life. Any remaining gaps might “not be more than 5%,” Raditya estimates.
“If we really want to strive for perfection, developing regional LLMs makes sense, especially since cultural nuances are difficult to master. But whether that justifies the extra investment remains to be seen,” Raditya tells Tech in Asia.
Making it more obvious
Do regional LLMs still matter?
Stay ahead in Asia’s tech landscape
This is premium content. Subscribe to read the full story.
The opportunity now lies in making end products, not competing with tech giants’ foundational models, industry insiders say.
We know this is not ideal. ⌛ Sign up in 20 seconds. Cancel anytime.
Our subscriber community includes professionals from these companies:
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.
