China Moved Its Data Centres Into Space. Here Is What Every Leader Needs to Know

Data centres now consume more electricity than the entire United Kingdom.

Every home. Every hospital. Every factory. And most of that energy becomes heat — vented into the sky, wasted, gone. Until very recently, that was simply accepted as the cost of doing business in the age of AI.

Two regions looked at that waste and decided enough was enough. One found a way to turn it into a community asset. The other moved computing off the planet entirely. Both decisions are happening right now, and understanding them will tell you more about the next twenty years of AI leadership than any product announcement you will see this year.

---

The Scale of the Problem

When people talk about data centre energy consumption, the numbers tend to blur past without landing.

So here is a comparison that makes it real. Global data centres consumed between 415 and 460 terawatt-hours of electricity in 2024 — close to two percent of all power used on the planet. The entire United Kingdom uses roughly 300 of those units per year, for everything — every home, every hospital, every motorway light. Data centres alone now consume more than that, globally, every single year. Carbon Credits

And that figure is accelerating. AI workloads are driving demand toward 1,000 terawatt-hours by 2026 — roughly equivalent to Japan’s total electricity consumption. The heat has to go somewhere. For most of the history of computing, it went into the sky.

---

Europe Turns Waste Into Warmth

The European response is practical, intelligent, and quietly significant.

At the Equinix PA10 data centre in Saint-Denis, Paris, server heat is captured and upgraded with heat pumps to supply roughly 10,000 MWh per year — enough to heat around 1,000 homes, or maintain the Olympic Aquatics Centre’s pools at 27 to 28 degrees Celsius throughout the year. Those were the pools used at the 2024 Paris Olympics. The heat came from servers processing data. Aquatherm

The scale grows from there. In Finland, Microsoft and energy utility Fortum are partnering to supply around 40 percent of district heating demand for 250,000 people using waste heat from new data centres — one of the world’s largest data centre heat recovery projects, scheduled to go live by 2026. Aquatherm

In the UK, the numbers are similarly real. Waste heat from a Gatwick Airport data centre is set to supply 46 GWh of heating to homes in Crawley. A separate scheme in London will provide heat to 9,000 homes in the Old Oak and Park Royal area. Colloide

I have spent over a decade recording expert interviews on where technology and leadership intersect. The moment a physics problem becomes a community heating contract is the moment you know a solution has crossed from theory into infrastructure.

The revised EU Energy Efficiency Directive now requires operators of data centres above 1 MW to technically assess waste heat utilisation, submit a cost-benefit analysis, and report key data to an EU database annually — with these requirements in force from October 2025. This is no longer voluntary. The direction is set. Aquatherm

---

China Moves the Problem Off the Planet

Europe’s response is intelligent. China’s response is extraordinary.

On 14 May 2025, China launched the first twelve satellites of its Three-Body Computing Constellation — an ambitious space-based supercomputing network led by Zhejiang Lab — aboard a Long March 2D rocket from the Jiuquan Satellite Launch Centre. The name is deliberate. It references Liu Cixin’s science fiction novel, The Three-Body Problem — a story about civilisations competing for survival across impossible distances. Business Standard

The numbers require some translation to land properly.

Your laptop handles roughly one billion operations per second. A high-end gaming machine manages perhaps ten billion. Each of these satellites handles 744 trillion operations per second — and all twelve are interlinked via laser communication running at 100 gigabits per second, working together as a single machine. That is 744,000 times more powerful than the fastest consumer hardware available today, operating in orbit, powered by near-constant solar energy, cooled by the cold of space. Business Standard

The ultimate goal is a constellation of 2,800 satellites delivering a combined real-time processing capacity of 1,000 peta operations per second. One peta is one thousand trillion. The full constellation would deliver one quintillion operations every second — in orbit, off the grid, beyond the reach of any terrestrial energy constraint. Editorialge

---

The Money Behind It

This is not a research programme. It is a capital commitment.

Orbital Chenguang, a Beijing-based startup, has obtained strategic credit lines totalling 57.7 billion yuan — approximately $8.4 billion — from twelve major financial institutions, including the Bank of China, the Agricultural Bank of China, and the Bank of Communications. SpaceNews

The long-term goal is a gigawatt-scale computing system in orbit by 2035, aligned directly with China’s 15th Five-Year Plan. This is not a moonshot. It is a line item in a national strategy document. NextBigFuture

When Elon Musk saw the SpaceNews report on Orbital Chenguang, he posted a single word on X: Interesting. That is not the response of someone unimpressed. That is the response of someone who recognises a serious move when he sees one.

---

What This Means for Leaders

There is a habit in boardrooms of treating AI as a software decision. Choose the model. Choose the vendor. Move on.

That framing is dangerously incomplete.

AI is an energy decision. The organisations that build their strategy around the energy equation — not just the capability one — will hold structural advantages that compound over years. They win on cost. They win on scale. They win on independence. And they win on ethics — because the organisation that does not depend on an overloaded grid is the one that keeps running when others cannot.

Global data centres emitted approximately 182 million tonnes of CO₂ in 2024. Some projections show the sector could generate up to 2.5 billion tonnes by 2030, driven by AI expansion. The organisations caught unprepared by that trajectory will face cost pressure, regulatory constraint, and reputational exposure at the same time. Carbon Credits

China is planning in decades. Europe is converting waste into warmth. Both are moving faster than most Western leadership teams are currently discussing.

---

Where the Real Advantage Lives

None of this means ground-based computing is finished. Satellite lifespans run to five or six years, launch costs remain significant, and the engineering complexity of scaling to gigawatt-class capacity in orbit is genuinely formidable. The economics of orbital computing are still being proven. TechRepublic

What it does mean is that the organisations positioning for the next decade are not simply asking which AI model to deploy. They are asking who controls the compute, who controls the power, and what happens when the grid cannot keep pace with demand.

China answered that question by looking upward. Europe answered it by routing warmth through the pipes of a quarter of a million homes.

Both answers reward the same quality in a leader: the willingness to look at a constraint and ask what becomes possible if you resolve it entirely — rather than manage it indefinitely.

That quality is what separates the leaders of the next era from those who simply adopted the tools of the last one.

The race is already underway. The energy equation is being solved. The only question remaining is whether your organisation is part of the answer.

---

Subscribe to Monday Influencer® — weekly expert signal for leaders who want to understand what is actually happening, not what is being announced. mondayinfluencer.com | First month $1, then $19.95/month

Read more on Substack: futureproofleader.substack.com

Free tools: natschooler.com/tools