The energy challenge of supercomputers overcoming all limits
Energy challenges and technological innovations are at the heart of the race for next-generation supercomputers
by Lab24
3' min read
3' min read
1000000000000000000000. A trillion (according to the long scale numerical naming system), a trillion trillion, 21 zeros. A number that is difficult to write down, describe or even imagine, but destined to become the new benchmark for supercomputer computing speed within the next five years.
Indeed, Japan is aiming, by 2030, to develop the world's first supercomputer with a speed of 1 zettaFLOPS, i.e. capable of performing one trillion calculations per second. A hitherto unexplored threshold.
The operational phase of the project officially began in recent weeks. On 18 June, the Japanese multinational Fujitsu Limited announced that it had obtained a contract from the Japanese research and development institute RIKEN to design the next-generation supercomputer. This initial basic design is scheduled to last until 27 February 2026.
The new supercomputer, provisionally called FugakuNEXT, represents the evolution of the Fugaku, currently seventh in the ranking of the world's most powerful supercomputers.
Topping the TOP500 list are three US supercomputers: El Capitan, Frontier and Aurora, installed at the US Department of Energy's laboratories. El Capitan can perform up to 2,746 million billion operations per second. As illustrated in Lab24's long form, if everyone on Earth were to perform one operation per second for a year, they would 'only' perform 258 million billion operations.
