The $600 million machine, capable of performing 2.79 quintillion calculations per second, will handle classified tasks to secure the US  nuclear stockpile and conduct cutting-edge simulations, including research using artificial intelligence. 

El Capitan, housed in a state-of-the-art data centre approximately 30 miles northeast of Silicon Valley, sits alongside a smaller sibling, Tuolumne, designed for non-classified work.

 While Tuolumne is just one-tenth the size of El Capitan, it is still powerful enough to rank 10th among the world’s most advanced supercomputers. The two machines, constructed by HPE and powered by AMD chips, are expected to significantly enhance research capabilities.

Livermore Lab deputy director Bradley Wallin said: "While we're still exploring the full role AI will play, there's no doubt that it is going to improve our ability to do research and development that we need.”

El Capitan's has 87 computer racks and associated infrastructure weigh 1.3 million pounds — roughly the weight of four blue whales — and it requires 30 megawatts of power to operate. This immense processing capability is equivalent to one million high-end smartphones working simultaneously.

For HPE, which acquired supercomputer maker Cray in 2019, El Capitan represents another milestone, building on its success with supercomputers at Oak Ridge and Argonne National Laboratories.

For AMD, it underscores the company’s bid to take on  Nvidia in the AI training market.

This pleased AMD CEO Lisa Su: “"I'm smiling from ear to ear," she said – which must have been quite scary to see.