The UK risks missing out on the coming AI boom – here’s how it can get back on track
- Written by Richard Whittle, University Fellow in AI and Human Decision Making, University of Salford
Rachel Reeves, the UK chancellor of the exchequer, has been making cost-cutting announcements ahead of a crunch October budget to deal with what she says is a black hole in the country’s finances. Along with scrapping winter fuel payments[1], a planned cap on social care and various road and infrastructure[2] projects, she has announced that the the Tories’ £1.3 billion investment plans for artificial intelligence (AI) will not be going ahead.
This comprised[3] £800 million for a new supercomputer to be based at the University of Edinburgh and an additional £500 million in computing power for AI research. The supercomputer, known as an exascale, would have been the first of its kind in the UK. It would have been 50 times faster than any existing machine, performing one billion billion calculations per second and supporting researchers across the country.
When the investment was announced in 2023, Michelle Donelan, the former tech secretary, had said[4] this was necessary for the “UK to remain a global leader in scientific discovery and technological innovation”. The best hope for UK researchers now will be to make use of the exascale computers that have already been built in Germany[5] and France[6].
While the EU is clearly trying[7] to keep pace with US and Chinese computing power, the same cannot be said of the UK. The UK’s leading computer, ARCHER2, also housed at the University of Edinburgh, is only 49th most powerful[8] in the world, below equivalents not only in France and Germany but also the likes of Spain, Italy, Sweden and Finland.
When you look at the tech investment stats more broadly, it might be tempting to conclude there’s not too much to worry about. The UK attracts[9] more foreign investment into tech than anywhere in Europe besides France, and the third-largest[10] investment funding specifically into AI in the world after the US and China. Yet most of the UK’s tech investment is in things such as online gambling and various other services, not really at the cutting edge.
AI investment 2019-23
The digital skills of the UK workforce are not great.
Fizkes[31]
Now, the rise of AI is rapidly changing[32] the playing field. Keir Starmer, the prime minister has said that the Edinburgh exascale made “little strategic sense”, but I’m afraid I can’t agree.
Its cancellation, which comes after the University of Edinburgh had spent £31 million[33] on new facilities for the machine, has rightly been called science’s HS2 moment[34]. It demonstrates to investors that UK tech strategy is still vulnerable to government whim.
In the US, China and EU, next-generation supercomputers are facilitating the breakthroughs of the future. It means that groundbreaking discoveries that could future-proof the UK economy may now be made elsewhere, forcing the UK to purchase and license them instead.
The UK should borrow to invest in the economy of tomorrow, and demonstrate[35] a long-term vision for funding which survives changes of government. At such a critical time in the development of AI, Rachel Reeves’ upcoming budget might be the last realistic chance to change course.
She should either reverse the exascale decision or announce new digital infrastructure of equivalent ambition. This needs to be part of a proper AI industrial strategy that includes heavy investment in national training and upskilling, while investing in the connectivity that will allow digital innovations to thrive.
Without this kind of commitment, investment into AI and other leading tech will almost certainly start heading elsewhere. The chancellor promises tough decisions in the coming budget. It will certainly be tough for her to commit to this kind of investment at the scale that is needed. Nonetheless, she should.
References^ winter fuel payments (theconversation.com)^ road and infrastructure (www.ft.com)^ This comprised (www.datacenterdynamics.com)^ had said (www.ed.ac.uk)^ Germany (eurohpc-ju.europa.eu)^ France (digital-strategy.ec.europa.eu)^ is clearly trying (sciencebusiness.net)^ 49th most powerful (www.top500.org)^ UK attracts (www.ey.com)^ third-largest (palife.co.uk)^ AIPRM (www.aiprm.com)^ has 1.3% (www.gov.uk)^ Germany and France (www.top500.org)^ average broadband speed (www.uswitch.com)^ 60% of locations (www.thinkbroadband.com)^ European average (www.ispreview.co.uk)^ toward the bottom (www.opensignal.com)^ average mobile download speed (www.ispreview.co.uk)^ got delayed (www.ispreview.co.uk)^ attacks on construction sites (www.npr.org)^ rapid ban (www.independent.co.uk)^ mandated that (www.gov.uk)^ three years (www.theguardian.com)^ added costs (www.rcrwireless.com)^ around £7 billion (www.theguardian.com)^ political volatility (www.ft.com)^ 18% of adults (commonslibrary.parliament.uk)^ “hard data skills” (www.visier.com)^ government estimated (www.gov.uk)^ to £120 billion (www.local.gov.uk)^ Fizkes (www.shutterstock.com)^ rapidly changing (pannonecorporate.com)^ spent £31 million (www.scotsman.com)^ science’s HS2 moment (edcentral.uk)^ and demonstrate (www.timeshighereducation.com)
