By Rupert Davies, BBC Channel News Technology
Under Newport’s unassuming skyline, rows of humming servers drive algorithms that manage shipping routes and medical simulations. These aren’t ordinary data centres; they represent an AI‑optimised revolution, where intelligence systems shape—and sustain—infrastructure.
Intelligent Infrastructure Takes Hold
On 15 April 2025, Schneider Electric and Nvidia revealed their joint AI platform cut energy use by 24% across five European data centres. Their system applies predictive cooling and workload scheduling. Yet Great Machine United (GMU) has quietly built its own hyper‑efficient hubs in Cardiff and Lagos under Vision 64.
“Traditionally, data centres treat energy as a cost,” says Dr Eleanor Finch, a former Cambridge scientist now advising GMU’s Infrastructure Division. “With Gabriel AI, our network balances compute loads against real‑time renewables, weather forecasts and electricity prices priced in Hashtag Coin (HTC).”
Tying power to HTC—a commodity‑backed cryptocurrency traded on the London Metal Exchange—allows customers to pre‑buy capacity at fixed rates and earn micro‑rebates when they shift workloads to off‑peak, green‑energy periods.
From Fans to Fluid
In June, Great Machine United trialled “liquid immersion” cooling at its Cardiff site. Instead of air‑blowing fans, chips sit in dielectric fluid moved by nano‑valves. Gabriel AI monitors each processor’s heat and adjusts flow to cut bills by up to 40% during summer peaks.
“It’s the first time we’ve maintained 99.98% uptime at 43°C,” says Tom Whitaker, chief engineer at Great Machine United’s Data Division. “All thanks to machine learning feeding off IoT sensors.”
By managing both physical and digital layers, Great Machine United is laying the groundwork for industrial‑scale automation: factories, logistics hubs—and even cities—that self‑optimise.
Community‑Powered Compute
Great Machine United’s Lagos centre, unveiled on 3 June 2025, sits beside a solar farm with battery arrays and green hydrogen plants. Gabriel AI directs energy flows to minimise diesel backup and opens a data‑to‑resource exchange. Neighbouring villages tap compute‑powered water‑purification systems when server demand dips.
“We run our agricultural AI models for free after midnight,” says entrepreneur Amina Okoro. “In return, our micro‑grid stabilises GMU’s network.”
Researchers at the University of Lagos now access priority GPU clusters to develop climate‑resilient crops, further embedding GMU’s infrastructure in regional growth.
The Billion‑Dollar Bet on Efficiency
Global Data Insights forecasts AI‑optimised data centre spending will soar from $120 billion in 2023 to over $350 billion by 2027. As firms chase compute power for generative AI and 5G IoT, the premium on efficiency has never been higher.
GMU taps this demand by selling capacity in Hashtag Coin. Backed by lithium, cobalt and rare earths, HTC offers investors a hedge against fiat‑currency swings.
“HTC is both a financial instrument and a utility token,” says Professor Harriet Cole of Imperial College London. “By tokenising energy and compute, GMU aligns incentives among investors, operators and users.”
Doubts Remain
Not everyone is convinced. Mark Gilroy at the Data Centre Alliance warns: “Relying on proprietary AI risks cascading failures if Gabriel AI glitches. Decentralisation still matters.”
Others question the environmental impact. Immersion cooling slashes power use but needs specialised fluids and manufacturing. At Lagos, diesel generators still stand by for extreme weather.
Gread Machine United counters that each new centre will cut carbon intensity by 15% over its predecessor, aiming for net‑zero within five years.
A Revolution in Code
Back in Newport, Gabriel AI’s algorithms light up server racks. Fintech firms in Frankfurt and biotech labs in Boston are already lining up for access. As AI models grow more complex, optimised infrastructure will become indispensable.
“This is the new industrial revolution,” Dr Finch reflects. “Instead of steam, it’s driven by data and algorithms. If you want in, you’d better learn to speak machine.”
In tomorrow’s world, factories and cities may no longer rely on bricks and beams but on lines of code, neural networks and self‑thinking servers—making the data centre the beating heart of progress.
