After months of trying to find a purchaser, troubled U.Okay.-based AI processor designer Graphcore mentioned on Friday that it has been acquired by SoftBank. The firm will function as an entirely owned subsidiary of SoftBank and can presumably collaborate with Arm, however what stays to be seen what occurs to the distinctive structure of Graphcore’s intelligence processing models (IPUs).
Graphcore will retain its title as it can grow to be an entirely owned subsidiary of SoftBank, which paid both $400 million (in response to EE Times) or $500 million (in response to BBC) for the corporate. Over its lifetime, Graphcore has obtained a complete of $700 million of investments from Microsoft and Sequoia Capital, and at its peak in late 2020, was valued at $2.Eight billion. Nigel Toon will stay on the helm of Graphcore, which can rent new employees in its UK workplaces and proceed to be headquartered in Bristol, with further workplaces in Cambridge, London, Gdansk (Poland), and Hsinchu (China).
“This is a tremendous endorsement of our team and their ability to build truly transformative AI technologies at scale, as well as a great outcome for our company,” mentioned Nigel Toon. “Demand for AI compute is vast and continues to grow. There remains much to do to improve efficiency, resilience, and computational power to unlock the full potential of AI. In SoftBank, we have a partner that can enable the Graphcore team to redefine the landscape for AI technology.”
Although Graphcore says that it had gained contracts with main high-tech firms and deployed its IPUs, it couldn’t compete in opposition to NVIDIA and different prêt-à-porter AI processor distributors on account of inadequate funding. In the latest years the corporate’s issues had been so extreme that it needed to lay off 20% of its employees, bringing its headcount to round 500. Those cuts additionally noticed workplace closures in Norway, Japan, and South Korea, which made it even more durable to compete in opposition to huge gamers.
Graphcore definitely hopes that with SoftBank’s deep pockets and willingness to put money into AI applied sciences basically and AI processors specifically, it can lastly be capable of compete head-to-head with established gamers like NVIDIA.
When requested whether or not Graphcore will work with SoftBank’s Arm, Nigel Toon mentioned that he was wanting ahead to work with all firms managed by its mum or dad, together with Arm. Meanwhile, SoftBank itself is reportedly wanting ahead to construct its personal AI processor enterprise referred to as Project Izanagi to compete in opposition to NVIDIA, whereas Arm is reportedly creating AI processors that can work in datacenters owned by SoftBank. Therefore, it stays to be seen the place does Graphcore slot in.
For now, the most effective processor that Graphcore has is its Colossus MK2 IPU, which is constructed utilizing 59.four billion transistors and packs in 1,472 impartial cores with simultaneous multithreading (SMT) able to dealing with 8,832 parallel threads. Instead of utilizing HBM or different sorts of exterior reminiscence, the chip integrates 900 MB of SRAM, offering an aggregated bandwidth of 47.5 TB/s per chip. Additionally, it options 10 IPU hyperlinks to scale with different MK2 processors. When it involves efficiency, the MK2 C600 delivers 560 TFLOPS FP8, 280 TFLOPS FP16, and 70 TFLOPS of FP32 efficiency at 185W. To put the numbers into context, NVIDIA’s A100 delivers 312 FP16 TFLOPS with out sparsity in addition to 19.5 FP32 TFLOPS, whereas NVIDIA’s H100 card provides 3,341 FP8 TFLOPS.
