There was an FT report this week that the world's largest technology companies are now committing to around $725 billion of investment in artificial intelligence infrastructure this year. That is an extraordinary sum to spend on what is, effectively, unproven technology that might be extraordinarily harmful to human well-being, that could undermine the fight against the consequences of human-created climate change.

It is also deeply revealing if contextualised, because organisations like the International Energy Agency and the United Nations continue to warn that the world is failing to invest at the scale required to tackle climate change, despite knowing exactly what needs to be done.

And as the House of Commons Library has reported, climate change is now recognised in international law as an “urgent and existential threat”, with states having clear obligations to act at the highest possible level of ambition. As they note, a 2025 advisory opinion from the International Court of Justice makes clear that governments must both reduce emissions and protect people from harm, with growing scope for legal challenge if they fail to do so.

Meanwhile, the UN currently suggests that emerging markets and developing countries, excluding China, need to spend between $2.3 trillion and $2.5 trillion a year by 2030 to meet essential climate goals. That is four times the current investment rate. Other UN reports estimate the total required annual investment at about $6 trillion per annum.

In that case, the contrast between the apparent ease of funding AI investment and the actual investment that society requires at the moment could not be more stark.

First, what this AI investment makes clear is that there is no shortage of money. $725 billion is not being scraped together with difficulty. It is not dependent on reluctant taxpayers. It is not being constrained by so-called “fiscal headroom”. It is being mobilised with apparent ease because large corporations believe there is profit to be made. This exposes a myth that dominates political debate: that we cannot “afford” to deal with climate change. Clearly, we can afford to do whatever we choose to prioritise.

Second, this makes clear that markets allocate capital according to expected private return, and not social need or necessity. Artificial intelligence promises potential for monopoly rent extraction, data control, and new, highly addictive revenue streams. Climate investment, by contrast, often produces public goods: clean air, stable temperatures, reduced risk. These are not easily captured as profit, even though they are essential for survival. So capital flows to AI rather than to the climate transition.

Third, urgency is not the issue here: direction is. The speed with which this level of funding has been committed demonstrates that, when the incentive exists, transformation can happen very quickly. There is no technical barrier to mobilisation. There is only a failure of political will to direct it.

There are at least three implications that follow from this.

The first is that the idea that governments must wait for growth to fund action on climate change is exposed as nonsense. If $725 billion can be created and deployed for AI because investors anticipate returns, then governments that issue their own currency can create money to fund climate investment when the need is overwhelming. The constraint is not money. It is real resources and political choice that are constraining the allocation of resources to this overwhelmingly critical task.

The second is that reliance on private capital to solve climate change is always misplaced. Private capital has made its preference clear. It will invest where returns are highest and easiest to capture. Climate mitigation often requires long-term, coordinated, system-wide change. That is precisely what markets are least well equipped to deliver.

The third is that the current system is misallocating resources despite the existence of existential risk.  We are not short of engineers, materials, or financial capacity. But we are choosing to deploy them to optimise advertising algorithms, automate existing processes, and consolidate corporate power, rather than to decarbonise energy systems, retrofit housing, or restore ecosystems. That is a political economy failure, not a technical one.

There are obvious responses that are required if we take this seriously.

Firstly, governments must lead the climate transition directly, with large-scale public investment in energy, housing, transport, and nature restoration. No apologies for doing so need be offered.

Secondly, regulation must change the investment landscape so that high-carbon activities are curtailed, and low-carbon alternatives are mandated. Private capital will follow when the rules change. It always does.

Thirdly, we have to recognise that not all investments should be prioritised equally. When resources are finite, choices have to be made, and some uses of capital are clearly more socially valuable than others. The climate transition must also be just, linking that action to employment, training, and social security, so that people benefit from change rather than fear it.

But what the juxtaposition of $725 billion for AI and inadequate funding for climate action ultimately reveals is something quite fundamental. We do not have an economic system that is designed to meet human needs. We have one designed to maximise private return. Until that changes, we will continue to see extraordinary sums mobilised at speed for activities that promise profit whilst causing potential harm, and persistent failure to act on issues that determine whether we have a habitable planet.

The question is not whether we can afford to tackle climate change. It is whether we are willing to redesign our economy so that we choose to, and can actually do so. That is what is really required.

PDF of article