In January, as business and government leaders descended on Davos, Switzerland, for this year’s meeting of the World Economic Forum, the prevailing wisdom was that American tech giants were winning the race for dominance in artificial intelligence while China, and indeed the whole of Asia, lagged behind.
Among the assembled cognoscenti, the outlook for China’s AI companies ranged from bleak to downright dismissive. The share prices of America’s “Magnificent Seven” hovered at all-time highs while Chinese tech stocks languished, dragged down by perceptions that Beijing’s heavy-handed regulation was strangling innovation. Meanwhile, Japan and South Korea weren’t generally considered leaders in AI, and the experts saw Southeast Asian economies as too small, too poor, and too fragmented to be anything more than bit players.
But the Davos crowd had barely adjourned before the West’s smug assumptions began to unravel. First came reports that DeepSeek, the AI arm of an obscure Hangzhou hedge fund, had developed a large language model, called R1, that matched the performance of OpenAI’s latest LLM. As Nicholas Gordon notes in his DeepSeek deep dive, R1 astonished not only by coming out of nowhere but because developers claimed to have trained it for a mere $6 million—a fraction of the tens of billions that Silicon Valley companies are spending on AI platforms.
Gordon argues that DeepSeek’s breakthrough matters because it not only shows Chinese companies’ ability to innovate in AI, it also suggests that the frontier of AI development is wide open to upstarts from almost anywhere.
The same week of R1’s release, a more familiar Asian player, Japanese tech investor Masayoshi Son, made a surprise appearance at the White House, where President Donald Trump announced that Son’s SoftBank Group would take the lead in bankrolling Stargate, a $500 billion partnership that aims to turbocharge American AI infrastructure. On April 1, SoftBank committed to invest up to $40 billion in OpenAI, valuing the AI pioneer at $300 billion. In a profile of Son, I explain how the 67-year-old billionaire—who has arguably made and lost larger fortunes than any investor in the history of capitalism—bounced back from the debacle of his Vision Fund to become one of AI’s most powerful players.
I also visit southern Malaysia to understand why Johor, the state just across a strait from Singapore, could emerge as one of the world’s biggest data center corridors. What’s new about data centers devoted to training LLMs is that unlike those used to run video games, operate e-commerce platforms, or handle financial transactions, they don’t need to be close to customers. They can be built wherever land, water, and—crucially—power are abundant and inexpensive. Malaysia is racing to provide those resources in a bid to propel itself to the forefront of the next technological wave.
Story Continues
