Hello Reader,
I'm now back in Singapore after attending CES 2026 in Las Vegas, a massive, incredible show of which I admittedly could only attend a small fraction of. I'll leave you to read all about it in the stories below, and I do plan to write a couple more enterprise and data centre-centric pieces next week.
For today, I want to talk about the growth of data centres in Southeast Asia.
Southeast Asia's moment
If you recall, I highlighted in an early-December commentary two possible scenarios to the current data centre boom: a bubble that's set to burst with catastrophic consequences, or the start of a supercycle unlike anything before.
I noted how Asia will need more data centres in either scenario, which means the region will be somewhat insulated even in the worst-case scenario. But as I rested during my year-end-break and talked to industry insiders in December, I'm now leaning towards a supercycle that is inching its way to Asia.
Specifically, I am hearing of an Eastern hyperscaler doubling down with plans to build significantly more data centre capacity in Southeast Asia. I've also had coffee with a senior executive of a large operator who admitted having more demand in their order books than they could fulfil. What a fascinating problem.
Sure, Nvidia's newfound ability to ship H200 GPUs to China might blunt some of the demand, notwithstanding its demand for full upfront payment with no option to cancel. But will Chinese hyperscalers with options really want to bet against a deliberately mercurial President Trump? Or that that Chinese government might change its mind again and impose a ban on its end?
More than just training
But what clinched it for me is the fact that AI is here to stay. You see, I was invited to attend CES by Lenovo, and it was during one of the briefings that Yuanqing Yang, Chairman and CEO of Lenovo, spoke of the different types of compute and how AI is merely the latest evolution.
In the closed-door session for media and analysts, Yang broke it down into traditional compute, the cloud, LLMs, and frontier AI models. That's right, Lenovo sees room beyond the global race to build ever-larger data centres like what Elon Musk's xAI is doing with the announcement of its third AI data centre this week.
In his view, there is ample room for Lenovo to compete in every one of these segments. Moreover, the current craze for building gigantic data centres to train frontier models is but one segment of the pie - there is plenty of work left to implement AI in businesses.
What I found fascinating is that rather than dividing everything into "AI training" or "AI inference," Yang cuts it lengthwise into personal AI, enterprise AI, embedded AI, and hybrid AI. And yes, Lenovo is well-positioned to jump in, whether in terms of personal computers, traditional and AI infrastructure, specialist AI models, or consultancy services.
The race to build data centres here is accelerating. I'm excited to see how 2026 shapes up. What are you hearing from your end? As usual, simply hit reply to this email and I'll get it in my inbox.
I look forward to hearing your thoughts.
Regards,
Paul Mah
Content collaboration
Love the way I write? I'm opening up for more content collaborations. If you are looking for event coverage, brand storytelling, or a way to share your unique perspectives with a broader audience, fill in the form
here and I'll get in touch.
