SoftBank Acquires Graphcore: A $600M Move in AI & Telecom

The Japanese conglomerate SoftBank announced the acquisition of the UK-based AI chipmaker Graphcore. While the precise financial terms weren’t disclosed, sources cited by the Financial Times valued the transaction at $600 million. Graphcore, 13 known for its advanced AI chips called intelligence processing units, has been instrumental in training and operating large language AI models.


Although Graphcore’s technology rivals that of industry giant Nvidia, the company has struggled to commercialize its products. In 2020, Graphcore was valued at $2.8 billion, but its recent performance has been disappointing. The firm reported only $2.7 million in sales in 2022—a 46% decline from the previous year. The pre-tax loss that year was a staggering $205 million.


Due to these struggles, Graphcore took cost-cutting steps in 2023, which included a 20% reduction in workforce and the closure of offices in Norway, Japan, and South Korea. The company faced “material uncertainty” about its survival and sought additional funding.


Under SoftBank, Graphcore expects to rejuvenate its operations. “Demand for AI compute is vast and continues to grow,” said Graphcore co-founder and CEO Nigel Toon. “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.”


SoftBank has significantly amplified its focus on AI over the past year. Founder Masayoshi Son stated that realising Artificial Superintelligence (ASI) was his primary focus, indicating a willingness to invest $9 billion annually in AI. This recent acquisition fits well with his ambitions for future large-scale

Read Also
Google Commits to Net Positive Water for Data Centers by 2030, Invests $17 Million in Water Projects
Eagle Rock Proposes $8.8 Billion, 1GW Data Center Campus in Illinois
Google and Intersect Power Break Ground on Co-located Data Center and 1GW+ Energy Project in Texas

Research