For centuries, a succession of raw materials defined which governments and economies grew – and which did not.
First, coal and steam, then, oil and electricity.
Each of those ages brought with it a period of profound change. Radically reshaping living standards and the labour markets of the time, with new jobs in different places. More money, and more things to spend it on.
Today, we find ourselves in the midst of another epochal shift. Its implications for our prosperity and our security will be no less seismic than those before.
Who swims – and who sinks – all depends on compute. Because, when it comes down to it, the AI era is no less material than any other.
The places and people who are shaping our economies have simply changed. Instead of collieries and oil wells, it’s the mines and refineries where silicon is processed.
It’s not the vast manufacturing plants of the past who dominate the stock exchange, it’s the companies who are designing ever more powerful chips and the businesses using them to train ever more powerful models.
I don’t want to underplay the significance of this change. To dismiss the economic consequences of the ‘rewiring’ we are witnessing in real time. But where the dynamics of the age of compute really differ, I think, is in the role of the state.
The state’s role in the economy has never been stable or predetermined. Each era poses the same questions of each government.
How to grow the economy? How to protect people? How to build better lives for our citizens?
Each time, the state must respond to those questions anew. Its legitimacy and longevity simply depend on it.
Today, though, these questions feel almost existential. The old answers just simply won’t cut it any more.
And the certainties we have depended on for decades are being swept away.
In the age of compute, we cannot – must not – be afraid to contemplate a sweeping change of course. That is what the UK’s AI Opportunities Action Plan sets out to do.
In the UK and the US, there are communities that have been left behind by the pace of change. Abandoned by industry, they are left clinging to the rusting remnants of the industrial age.
Losing faith in governments that have failed to deliver promise after promise and failed to deliver rebirth and renewal. I understand why people in these places worry that AI will not be working for them.
That, as start-ups in Silicon Valley and London create wealth and prosperity for some, the rest of the economy will remain just as stagnant and unproductive as before. But I don’t believe there’s anything inevitable about that story.
In empty factories and abandoned mines, in derelict sites and unused power supplies, I see the places where we can begin to build a new economic model.
A model completely rewired around the immense power of artificial intelligence, where, faced with that power, the state is neither a blocker nor a shirker – but an agile, proactive partner.
In Britain, we want to turn the relics of economic eras past into AI Growth Zones.
With access to large power connections and a permissive planning system designed to cut the time it takes to start construction, these are the places where we’ll work with industry and local government to build compute infrastructure on a scale that our country has never seen before.
There is a real hunger for investment in Britain. People who are optimistic about the future, and hopeful for the opportunities which AI will bring them and deliver for their families and communities.
Earlier this year, we asked local leaders across the country to come to us with proposals for Growth Zones and how it could impact their areas. Since then, we’ve had over two hundred responses.
That is evidence of the ambition and appetite you can find in equal measure at the top of government in Britain right down to the grassroots of communities across the United Kingdom.
Today, I can announce that the responses we’ve received include several sites that could host very powerful data centres.
One of those sites will get close to 2 GW. In our former industrial heartlands, hundreds of acres of flat land are sitting completely unused and ready for construction.
Soon, though, this could be home to the largest data centre in Europe. And we have no time to waste. I want shovels in the ground this year.
Because, if states are to secure their sovereign role in the future of this technology, they simply cannot afford to wait. And we will not.
In the age of compute, we must offer more than just a place to invest. That’s why our AI Growth Zones will be the anchor for a more ambitious project. A project designed to unleash a new age of growth and prosperity across our nation, and build a smaller, smarter state.
One that is ready for the century to come.
Home to Nobel Prize winners like Sir Demis Hassabis, the U.K. has world-leading scientific capability in the development and deployment of AI. With a cradle-to-grave health service that has been running for 75 years, we also have uniquely rich data sets you cannot find anywhere else in the world.
And we have a government with the capacity and the political will to deploy transformative technology in every part of our public sector, from courtrooms and classrooms to hospitals and job centres.
Because we know that, if we want to deliver better services for citizens and better value for taxpayers, we have no other choice. In a country whose language and legal system are used around the world, that unique contribution – of global talent, data, and political will – can yield extraordinary results.
Today, every single stroke centre in England is using AI to interpret acute stroke brain scans and support doctors to make decisions about treatment. Early data shows this is cutting the time it takes to get patients in and out the door from 340 to 79 minutes.
[The incorrect figure was given in the speech as delivered. 140 minutes is the correct figure.]
And it’s tripling the chance of independent living following a stroke.
It’s something of a truism that compute is only as good as the people who are using it, and the data they put in it.
In Britain, we have both of those things in abundance. But the AI Opportunities Action Plan offers something else, too. A chance to test the models you are training in a country that is crying out for reform, and with a government ready to use AI to take on the great challenges that will define the century to come.
Tackling those challenges will require more than brute capacity. Building bigger or faster is not enough.
In the age of compute, states must build smarter, too. That’s part of the reason I’m here in San Jose.
Just around the corner at Lawrence Livermore, scientists are using El Cap – the world’s most powerful supercomputer – to advance the safety, security and reliability of your nuclear arsenal.
At Oak Ridge, they’re using Frontier to model stellar explosions, neutrino physics experiments and global climate patterns.
The US model of national labs shows what states can achieve by investing in world-class research infrastructure.
The strategic advantage it provides is unparalleled.
It won’t surprise you to know that I want to replicate that success in the UK.
Because I believe government has not just a role to play, but a responsibility to shoulder in ensuring that AI delivers better lives for all of its citizens.
And we cannot fulfil this responsibility without publicly accessible compute.
In our Action Plan, we are committed to increasing our public compute capacity by twenty fold by 2030.
And last year, Isambard, the first phase machine of our AI Research Resource, came online.
Built using Nvidia chips, it is named after Isambard Brunel – the engineer who built the British ships and railways that changed the age of steam forever.
Our scientists are already using it for protein mapping to deepen their understanding of heart disease – the leading cause of death globally.
If we want to make our economies strong again, our countries healthy and our citizens safe, ambitious, rigorous research will be critical.
States owe it to their citizens to support it. Not through diktat or directive, but through partnership.
That’s why, last week, we opened market engagement for the private partnerships we will need to deliver our public compute ambitions.
If you want to work together, I urge you to get in touch.
I spoke earlier about the big questions that all states must answer in the age of compute. About how to ensure that technical progress translates into prosperity. How to protect our national security in a new global economy. The question of research, and how states should support it, can be added to that list.
But there is another big question which we must confront. That is the question of energy.
Because, in this respect at least, the age of compute is no different from any other. Power – and its availability – will shape it indelibly.
I reject the doomers who claim that the energy demands of AI undermine the promise that this technology somehow possess. They were wrong before and they’re wrong now.
The very existence of the GPU defies what were once believed to be the limits of scientific possibility.
In the decades since, those limits have been defied again and again.
So there is no reason why the challenge of energy efficiency should be somehow insurmountable. Together, we have already made impressive progress.
NVIDIA’s Blackwell architecture – backed by processors designed by Arm – uses 25 times less energy than previous generations, and Isambard AI is the fourth most energy-efficient supercomputer in the world.
The real challenge, I think, is to ensure that innovation is not left behind in the race for scale.
To ensure that – even as we invest billions in compute infrastructure – we do not fail to challenge the tried and tested ways of delivering it. You don’t need me to tell you that.
You are the people who are pushing against the frontiers of energy efficiency – rethinking architectures, rethinking cooling systems and energy sources.
I mention energy, though, because I believe that states can be partners in that progress.
And I want the UK to be a laboratory for change.
A place where pioneers can challenge old orthodoxies.
Where they can achieve the impossible and set a new course for the age of compute.
Today, that project feels more urgent than ever before.
In the last few months, we have witnessed the emergence of a new ‘scaling law’ in AI. A law that – some argue – will make compute less important than it was before. I couldn’t disagree more.
Test-time scaling offers a complement – not a replacement – to pre- and post-training scaling methods.
An opportunity to use the compute we do have to unlock deeper forms of intelligence.
But it does not reduce in the slightest the critical significance of compute for states looking ahead to the century to come.
The age of compute isn’t going anywhere.
Without compute, no economy can thrive. No country can protect its people. No government can retain the trust of its citizens.
AI will bring deep disruption to almost every aspect of life as we know it. The logic of our economies and the legitimacy of the state are at stake.
Britain stands ready not just to face that disruption, but to embrace it with you.
Time and time again, we have worked together to shape a shared future, anchored in freedom, fairness, and the rule of law.
Government with government, business with business, researcher with researcher. This is an alliance whose breadth and depth have no parallel.
Today, we are the two foremost AI nations of the democratic world, and that alliance matters more than ever.
Britain is full of talented, forward-thinking people. People who are ready to throw off the shackles of caution and conservatism and seize the once-in-a-generation opportunity that AI offers.
With a government that is ready to get behind them. Ours is a country that is ready for investment, and ready for change.
I have talked a lot about collaboration already today, because, when it comes down to it, that is what I have come here to offer.
Not just an opportunity to invest in Britain but a chance to form a new kind of partnership.
A partnership that is tailored to the needs of our economic era.
That partnership does not shy away from wealth creation but embraces it, because we know just how much our citizens stand to gain.
It is rooted in a recognition of AI’s power to transform our economies – and a willingness to do what is necessary to make that transformation happen.
And it is anchored in the values we share – because a future without them is simply unthinkable.
This, I believe, is how the state survives in the decades to come.
Not through retreat or withdrawal.
Nor by rushing towards excessive rules and regulations that will stifle innovation and growth.
But through strategic, purposeful partnership with you – the protagonists of the age of compute.
Thank you.