Geography built it.
The minister cut the ribbon.

My feed this week: a property developer selling Sepang land to a data-centre developer for RM617.9 million cash, and a viral post celebrating Malaysia as one of only four net exporters of AI-related hardware, RM144.4 billion in approved data-centre investment behind it. Genuinely big numbers. Now the honest question nobody at the ribbon asks: which part of this did any ministry actually build?

What actually sold Malaysia

Ask the site selectors why the hyperscalers came. The answers, in order: Peninsular Malaysia sits outside the Ring of Fire — no earthquakes. Outside the typhoon belt — no seasonal shutdown risk. Flat land next to the world’s densest network hub in Singapore, at a fraction of Singapore’s cost. Power that is still among the cheaper options in the region.

Not one item on that list was created by a policy. I made this point in my data-centre economics piece: cheap power shortens the payback window, but geology protects it — and that combination is Malaysia’s real pitch. A knocked-out asset never repays a two-year window, so boring, stable ground is worth billions. Nature wrote the FDI brochure. The ministry printed it.

The scorecard reflex

To be fair before I swing: government did do real work — approvals moved fast, land got zoned, the grid signed the deals. Enabling is work. But watch the framing: every megawatt lands on a scorecard as if it were policy genius, every groundbreaking gets its row of scissors. It is the same reflex I keep writing about — a scoreboard that counts announcements. An advantage you inherited is being presented as an advantage you engineered. That confusion matters, because inherited advantages teach you nothing about keeping them.

Meanwhile, the frontier left the land

Here is what the ribbon-cutting misses. While we celebrate hosting the current generation of data centres, the people who design them are already moving to the next terrain.

China put a commercial data centre under the sea. Not a lab toy — Hailanyun’s cluster sits 35 metres below the surface off Shanghai: roughly US$226 million, about 800 AI servers, powered by offshore wind, cutting electricity use by at least 30% because the ocean is the cooling system. The pilot ran in Hainan in December 2022; commercial service took less than 30 months. Microsoft tested the same idea with Project Natick years earlier — and never dared to commercialise it.

And above us: Musk says SpaceX “will be doing” orbital data centres, built on Starlink V3 satellites launching from 2026. Jensen Huang backs the concept. Google’s Project Suncatcher plans TPU satellites with a prototype around 2027. A startup called Starcloud has filed for an 88,000-satellite constellation. In orbit, solar power is uninterrupted and cooling is free vacuum — the two costs that define this industry both trend toward zero.

Why I’d bet on China for the space round

My honest guess — and I will mark it clearly as a guess: China may win the orbital round, for a boring logistical reason. It is already launching the infrastructure. First 6G test satellite in orbit, space-to-ground laser transmission demonstrated at 6G standard, national 6G trials in phase two, two mega-constellations under deployment. When your country is already mass-launching satellites with high-bandwidth laser links, a data centre in space is not a moonshot. It is a payload change.

The US holds the chip advantage; China holds the deployment machine. That race is genuinely open — but only one side already treats orbit as routine freight.

To be fair about the leftovers

Now the necessary honesty on my own argument: leftovers still pay. The current wave brings construction, jobs, tax, ecosystem — Malaysia absolutely should take it, and taking it well is not nothing. Vietnam would trade places tomorrow.

The sin is not hosting. The sin is mistaking hosting for winning — believing the scorecard instead of studying the map. The same lesson always follows: advantages tied to one era’s playbook expire when the playbook does. If the next generation of compute lives underwater and in orbit, the question for Malaysia is not how many ribbons were cut this quarter. It is: what did we build on top of the wave while we had it — operators, skills, energy strategy, ownership — that survives the wave moving on? My earlier warning stands: ask who owns the chips, and who merely rents out the ground beneath them.

The bottom line

The business lesson is personal, not political. Run the same audit on your own company: how much of your advantage is inherited — the location, the licence, the old relationship — and how much did you build?

The inherited part is exactly the part the next wave routes around. Build the part that travels.