The US has officially lost its crown after China secretly built the world's fastest supercomputer
China has dealt the US a major blow in the global supercomputer race, ending America’s run at the top of one of the most closely watched technology rankings in the world.
Whilst it sounds quite niche, compared to all the bigger things going on in the world right now, it’s a monumental shift nevertheless. Supercomputers are used for some of the most demanding work in science, defence, medicine, climate modelling and artificial intelligence, making their speed a closely watched measure of national technological strength.
The US has particularly continued to dominate much of the high-performance computing conversation in recent years, especially through machines housed at government-backed national laboratories.
However, as reported first by outlets like The Guardian, the latest ranking shows that China has now moved ahead with a system that was not even previously listed among the world’s fastest publicly verified machines.
Advert
That machine is LineShine, a supercomputer based in Shenzhen, which has debuted at number one on the latest TOP500 list.
LineShine displaced El Capitan, the US supercomputer based at Lawrence Livermore National Laboratory in California, which has now dropped to second place.
According to scientists behind the TOP500 project, LineShine achieved 2.198 exaflops. That means it can carry out more than 2 quintillion calculations per second.
The result marks the first time since 2017 that a Chinese machine has topped the ranking, which is often seen as a measure of a country’s technological power.
El Capitan is now followed by two other US supercomputers, housed at national laboratories in Tennessee and Illinois.
Germany’s Jupiter machine has fallen to fifth place, with those five computers currently standing as the only publicly verified exascale systems in the world.
One of the most striking details about LineShine is the type of hardware it uses. This is because, unlike many modern high-performance computers, particularly those linked to artificial intelligence, LineShine runs entirely on conventional computer chips, known as CPUs.
That sets it apart from the GPU-heavy systems that have become central to the current AI race, with graphics processors now widely used to train and run advanced models — taking up the majority of supply chains and driving up memory-based components’ prices as a result.
However, the system’s performance comes with a large energy demand. TOP500 lists LineShine’s operating power at around 42.2 megawatts, underlining the huge infrastructure required to keep machines of this scale running.
The wider ranking also shows the gap between the leading machines and much of the rest of the world. The UK, for example, has 11 systems in the TOP500, with the University of Bristol’s Isambard-AI the highest-ranked among them at number 11.