Exascale Computing Development Confirmed for Edinburgh

Few now remember Black Arrow, UK’s 1970 satellite launcher. This developed out of the successful earlier Blue Streak rocket program. But when its first flight failed to reach orbit, the politicians of the day used that as an excuse to cut the project. I did work, though – a later rocket did actually launch a satellite called Prospero. The story is that the cancellation was decided after the rocket and satellite were ready to launch – and permission was given to go ahead as a means of disposing of it! Three rockets remains on display in Edinburgh, Liverpool and Woomera.

Prospero was officially deactivated in 1996 but was occasionally symbolically switched back on on its launch anniversary. It was last heard on 137.560 MHz in 2004. It is expected to remain in orbit until 2070.

Thus UK gave up on space.

Last year, as reported in ‘The Economist’, the Labour government seemed to be thinking of doing the same thing with the Archer programme based in Penicuik, in the Pentland Hills. The Archer2 is UK’s most powerful current system and cost £79 million to develop. However an exascale computer would be capable of processing a quintillion calculations per second, making it 50 times faster.

The previous government had allocated £1.3bn for technical and AI development, of which £800 million had been earmarked for the project at the University of Edinburgh. This would radically reduce the time taken to perform complex research, according to the director of the university’s supercomputer centre, and would be critical to the adoption of AI in the UK.

In August 2024 government sources were expressing the concern that too much of the total allocation would be going to Edinburgh and would draw up a national “AI roadmap”. This was despite the University of Edinburgh having already invested £31 million in the supercomputer hub, with a power supply to match.

Now, after a year’s delay, the University of Edinburgh is to become the UK’s first national Computing Centre and nearly all of the original funding (£750 million) will be made available to them.

But, in a fast-developing field like AI, this delay will have been damaging – and of course giving US competition time to catch up. It will inevitably have led to some talented engineers moving away.

The Labour government has belatedly endorsed the decision of its predecessor. But at one point it looked dangerously like becoming a re-run of the historic Black Arrow fiasco. Let us hope UK is still not too late to protect its national interests in AI development.

(this posting has relied heavily on the ‘Economist’ article entitled “It doesn’t compute”)