Fresh from having conquered the mobile market, U.K.-based microchip designer ARM Holdings Ltd. is now setting its sights on powering the world’s fastest computers. It’s just unveiled a new chip design for supercomputers, an extremely lucrative market where until now, it’s had zero presence. The new design, which is based on ARM’s mobile chips, has some substantial tweaks to provide a significant boost in computing power.

ARM’s announcement comes just weeks after the company agreed to be sold to Japanese telecoms giant SoftBank Group Corp. in a massive $32 billion deal. ARM is likely to use some of that cash to further its ambitions in the data center and Internet of Things (IoT) spaces.

The company’s latest design, which was unveiled at the Hot Chips conference in Cupertino, California, yesterday, also sends a signal to rivals like Intel, IBM and others that it’s just as capable of developing blazing-fast supercomputing chips as they are.

Japan’s Fujitsu Ltd. will build the first supercomputer packing the new chips, called the “Post-K supercomputer,” ARM said. Fujitsu recently said it would be dropping its own SPARC architecture in favor of ARM’s designs, and helped with the design of the new chip.

Japan, along with the U.S. and China, is racing to be the first nation to build a supercomputer that delivers 1 exaflop of performance, or one million, trillion calculations per second. Companies including Intel, IBM and NVIDIA Corp. also aspire to the same goal with their own chip designs.

The Post-K supercomputer will be the successor to Fujitsu’s K Computer, which is ranked as the fifth fastest computer in the world at present. Post-K will be between 50 and 100 times faster than the latter, which delivers 10.5 petaflops of peak performance using Fujitsu’s SPARC 64 VIIIfx chips.

ARM’s new design is based on the company’s 64-bit ARM-v8A architecture, and one of its interesting features is it relies on vector processor extensions to deliver a performance boost, much like the earliest supercomputer chips in the late 1980s and early 1990s did. Vector processors later went out of fashion as supercomputer designers looked to IBM’s more affordable RISC chips in the mid-1990s, before later switching to standard x86 processors. But just like any good fashion trend, history is repeating itself and vector processing is in vogue once again, with new designs like ARM’s and Intel’s Xeon Phi chips both sporting them.

ARM believes its new chip design will provide both a performance boost and also consume less power – an important consideration given the phenomenal rate at which supercomputing speeds are growing.

ARM isn’t the only upstart looking to disrupt the supercomputing space. China’s Sunway TaihuLight, currently the world’s fastest supercomputer, boasts its own homegrown ShenWei processor, which offers a peak performance of 125.4 petaflops.

Perhaps more importantly for ARM, the new design will help give it a leg up in the data center chip market, an arena where its so far struggled to make any inroads against the all-powerful Intel. ARM-based servers have been on the market for some time, but adoption remains low. However, with large server clusters quickly becoming the norm for emerging technologies like machine learning, these could clearly benefit from the low-precision calculations that a large congregation of ARM’s chips could provide.

Images via ARM Holdings Ltd. & Fujitsu Ltd.