AI is advancing faster than data centers can keep up in Asia, says a Groq exec

Business Insider

28-05-2025

As the world scrambles to scale artificial intelligence, Asia is running into a very physical problem: not enough data centers and infrastructure, said Ian Andrews, the chief revenue officer at semiconductor startup Groq.
There's a "huge challenge getting enough compute" for what the company wants to do in Asia, Andrews said on Wednesday during a panel at Singapore tech conference ATxSummit.
"It's a big problem that we have to go tackle," he added.
While most AI companies have focused on training large language models, Groq is betting on speed, building its own chips to run models faster. The gambit is that as AI models get better, inference — where the AI makes decisions or answers questions — will demand more computing power than training will.
The region is already seeing bottlenecks when it comes to infrastructure like data centers and power, Andrews said, adding that it's likely to worsen as AI becomes more widely adopted in the region.
"Keep in mind, we're still in the infancy of AI," he said.
Andrews also said that over the next five years, all applications might be driven by AI.
"There is no model in which we have enough data center capacity, enough power, and enough infrastructure to run all of that in this region," he added.
Model progression is a more solvable problem than the infrastructure ones, Andrews said.
Andrews' comments come as AI heavyweights like OpenAI deepen their footprint in Asia and governments across the region ramp up spending on infrastructure to support the technology.
OpenAI on Monday said it will soon set up an office in South Korea, its third in Asia.
The company's chief strategy officer, Jason Kwon, said growth in ChatGPT's user base in South Korea has been "off the charts." South Korea has the highest number of paid ChatGPT subscribers outside the US, he said.
In November, Taiwan's science and technology minister said that the government would spend $3 billion over three years to ramp up AI data centers and their computing ability.
Meanwhile, big tech companies are pouring hundreds of billions of dollars into global infrastructure to ramp up AI development.
Early ChatGPT was a 'toy'
To grasp how quickly AI is evolving, Andrews pointed to ChatGPT.
"It was only about 30 months ago that ChatGPT launched," Andrews said. "If you go back and look at what you can do with ChatGPT compared to where we are today, it was a toy," he added.
There's an "acceleration" in AI's capability, Andrews said.
"In the first quarter of 2025 alone, there were more state-of-the-art, frontier models launched than all of 2024," he said.
"I'm an accelerationist on this point. Things are going to move quicker than we expect in terms of the capabilities," he added.