How tech giants can scale AI without breaking their climate commitments

The artificial intelligence (AI) boom is running up a carbon debt.

By 2030, global data centre electricity use will double to nearly 1,000 terawatt-hours – more than the world’s fifth largest economy, Japan, consumes in a year. Clean power is coming online at record speed, but not fast enough to meet such runaway energy demand.

And there’s no slowing the AI arms race, so when clean energy can’t keep up, fossil fuels will likely make up the gap. The International Energy Agency (IEA) expects gas, oil and even coal to provide more than half of data centre energy use through 2030, generating over 300 million tonnes of CO₂ emissions annually.

The hyperscalers driving this transformation have each set ambitious net-zero targets of 2030 for Microsoft, Google and Meta, and 2040 for Amazon. But tech giants' emissions are still rising with barely four years to go until these first deadlines kick in. Every tonne emitted now adds to a mounting carbon debt that must be repaid to reach net zero.

The tension between soaring AI power demand, limited clean energy supply and urgent climate goals is now one of the biggest challenges in tech.

No company illustrates this more clearly than Microsoft. It is investing $80 billion in data centres during 2025, has signed clean energy deals in 24 countries and is betting big on both traditional nuclear and emerging fusion technologies. But its latest sustainability report shows its energy use has jumped 168% since 2020 and its total emissions are still rising.

As Melanie Nakagawa, the company’s chief sustainability officer, put it when discussing Microsoft’s 2030 carbon-negative moonshot: “the moon has gotten further away”. And even the most ambitious tech companies are struggling to keep up.

Balancing the net-zero equation requires steep emissions reductions, as well as carbon dioxide removal (CDR) solutions. These technologies pull CO₂ out of the atmosphere and store it for centuries or longer in soils, plants, oceans or underground.

Traditional avoidance offsets shift emissions from one party to another, but they remain in the atmosphere, warming the planet. In contrast, carbon removal deals with what’s already in the air, resulting in less carbon than before. The Intergovernmental Panel on Climate Change says we need to remove 5–10 billion tonnes of CO₂ per year by 2050. There’s no net-zero scenario without it.

Microsoft knows this and, just as it has acknowledged the challenge, it's also paving the way forward. The company has so far purchased more than 30 million tonnes of high-quality carbon removal. But the gap from megatonne to gigatonne scale is still enormous. Microsoft accounts for roughly 80% of the global market as of October 2025. To scale AI without blowing past climate commitments, other hyperscalers must match Microsoft’s ambition.

Why would a tech company invest billions in carbon removal right now? It's less about corporate responsibility and more about business resilience. Climate change is straining infrastructure, inflating insurance and capital costs, and disrupting global supply chains.

Carbon removal is one way to insulate growth from those systemic risks. And even if US policy-makers retreat from climate leadership, global companies know that regulation is a matter of when, not if – the UK and EU emissions trading schemes will soon integrate carbon removal into their compliance markets, for example.

Companies that act now will gain first-mover advantage in securing high-quality CDR supply. Those that wait for regulation to catch up risk supply shocks, higher costs and public scrutiny. The earlier companies invest in carbon removal, the lower the long-term cost for the business.

And signing long-term contracts also gives carbon removal companies the confidence to build, unlock financing and bring down costs, just as tech companies helped scale renewable power a decade ago.

The companies driving the AI buildout are no longer acting like large corporations, they’re operating at nation scale, reshaping global markets for electricity, materials and carbon itself. But private sector leadership alone will not be enough.

Governments must step up with clear rules, public procurement programmes and investment in infrastructure to move, measure and store carbon – just as they did in the past for renewables, broadband and vaccines. That means funding demonstration projects, accelerating permitting for storage sites and integrating removal into compliance markets, not just voluntary ones. Without early public scaffolding, the private sector cannot absorb the risk or move fast enough on carbon removal.

The AI boom is a once-in-a-generation opportunity to align private capital and public policy to tackle the most urgent engineering challenge of our time. AI has redrawn the climate curve, but carbon removal provides a way to pay down the carbon debt and keep innovation compatible with a liveable planet.