Inside F1's pop-up data centre in Singapore
A week to build, race on, and tear it all down.
F1 builds a data centre in Singapore in one week, races on it, then tears it down.
It's Formula 1 season in Singapore, and I had the chance to watch the Singapore F1 pre-race and visit the Event Technical Centre. What surprised me most were the parallels with modern data centres. It turns out the latest trends in the industry are things F1 has been doing for years.
Build me a data centre
At a briefing at the F1 Event Technical Centre, I was surprised by the sheer amount of digital infrastructure needed to give fans the best experience possible. It comprises dozens of kilometres of fibre optics, multi-path network redundancy, a hundred cameras, and an on-site network operations centre (NOC), all assembled at extraordinary speed.
Incredibly, everything is shipped in, deployed, and wired up within a single week. Nicholas Leong would probably say this is part of "the race" behind the race.
While I wasn't allowed to see the server racks and networking gear, I learned that F1 ensures resilience through multiple redundant paths, including the use of satellites. Crucially, switchover is near-instantaneous, lasting at most six seconds, during which replays are seamlessly slotted in by the main centre at Biggin Hill in the UK.
I counted at least 25 seats in the mobile NOC. I couldn't make out most of the systems on display; suffice to say I am somewhat inspired to upgrade from my current 40-inch work monitors.
The future of AI is hybrid
The evening wasn't only about the race infrastructure. I also had the chance to speak with Fan Ho of Lenovo, the tech giant that furnished the systems used by F1, and gleaned some insights about the AI rollout in the region.
The first is that despite productivity gains and surging use of AI tokens, only two or three out of 10 companies know where their tokens actually go. This dovetails with my own conversations with executives of large organisations. No wonder Lenovo wants to do the token-governance and control layer.
The second is that manufacturing and logistics are first to deploy AI, tackling challenges such as labour cost, labour shortages, and mundane work. But others such as healthcare are not far behind.
Finally, if you think about it, there are roles where robots, or physical AI, can work better than humans. For instance, robots with cameras leave traceable records that can be audited, and they can also see in wavelengths humans can't.
I came for the race and left thinking about data centres and failover times. Not what I expected from an evening at Marina Bay, for sure.
