SoftBank Group has injected more than $450 million into British AI chipmaker Graphcore, boosting founder Masayoshi Son’s increasingly aggressive bet that control over artificial-intelligence infrastructure will define the next era of global technology dominance.
A filing with the U.K.’s Companies House showed Graphcore issued a single share valued at roughly $457 million on April 10. A company spokesperson confirmed the funding came from SoftBank, according to CNBC.
The investment signals that SoftBank is not treating Graphcore as a distressed acquisition to be quietly absorbed, but rather as a pillar in a much broader effort to build an end-to-end AI empire spanning chips, robotics, data centers, cloud infrastructure, and artificial general intelligence, or AGI.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
The latest funding round also underscores how the global AI race is rapidly shifting from software models alone toward the far costlier battle for compute infrastructure and semiconductor control.
For SoftBank, the investment is part of a years-long transformation under Son, who increasingly sees AI hardware as the foundation upon which future economic and geopolitical power will rest.
“At the time of the acquisition,” SoftBank said Graphcore would collaborate with the company on developing artificial general intelligence, “when AI matches or surpasses human intelligence.”
The new capital appears designed to accelerate that ambition. A person familiar with the arrangement told CNBC the injection represents only “a portion” of the funding Graphcore is expected to receive from SoftBank this year, suggesting further investment could follow.
The move comes as global technology giants race to secure AI chip capacity amid unprecedented demand for computing power driven by large language models and AI agents. That race has largely been dominated by Nvidia, whose GPUs became the backbone of the modern AI boom. But soaring demand, supply bottlenecks, and concerns over dependence on a single supplier have pushed governments and corporations to search for alternatives.
Graphcore was once viewed as one of the strongest challengers to Nvidia. The company attracted hundreds of millions of dollars from investors, including Microsoft and Sequoia Capital, after developing its Intelligence Processing Unit, or IPU, architecture aimed specifically at AI workloads.
At one point, Graphcore’s valuation exceeded $2.5 billion. Yet the company struggled commercially as Nvidia tightened its grip on the market through its software ecosystem, CUDA developer tools, and expanding dominance in AI infrastructure.
Many AI customers preferred Nvidia’s mature ecosystem over smaller rivals with less software compatibility and limited scale. That left Graphcore squeezed financially before SoftBank acquired it in 2024.
Under SoftBank, however, Graphcore appears to have gained a second life. The Japanese conglomerate is increasingly assembling a vertically integrated AI infrastructure ecosystem. Its holdings already include Arm Holdings, the British chip designer whose architecture powers most smartphones globally and is becoming increasingly important in AI computing.
SoftBank acquired Arm in 2016 before relisting it on Nasdaq in 2023. The company also acquired silicon-design firm Ampere Computing in 2025, further expanding its semiconductor footprint. Together, those assets give SoftBank exposure across multiple layers of the AI hardware stack: chip design, processor architecture, AI accelerators, and data-centre infrastructure.
The strategy reflects Son’s long-held belief that AI will trigger an industrial transformation comparable to the internet revolution, but one requiring vastly larger physical infrastructure investments.
That thesis has become increasingly mainstream. Building and running advanced AI systems now requires enormous amounts of electricity, specialized semiconductors, networking hardware, and data-center capacity.
As AI models become more powerful, infrastructure costs are rising exponentially. That is why companies are now spending tens or even hundreds of billions of dollars securing computing resources.
SoftBank has emerged as one of the biggest financiers of that shift. The company is involved in the massive $500 billion Stargate AI infrastructure initiative alongside OpenAI and Oracle. It has also invested heavily in OpenAI itself and is reportedly exploring major AI data-centre projects in Europe.
Bloomberg reported Monday that SoftBank is discussing a large AI data-center initiative in France. The Financial Times separately reported in April that SoftBank plans to spin off and publicly list a standalone AI and robotics company in the United States, potentially as early as this year.
Together, the investments suggest Son is attempting to position SoftBank at the center of the AI infrastructure economy rather than merely backing applications built on top of it. That marks a notable shift from SoftBank’s earlier strategy through its Vision Fund, which spread capital broadly across consumer technology startups, many of which later struggled.
The company, with renewed focus, is now concentrating much more heavily on foundational AI infrastructure, which comes with geopolitical impact. Governments increasingly view semiconductors and AI infrastructure as strategic national assets tied to economic competitiveness and national security.
The United States, China, Europe, and Japan are all investing heavily to reduce dependence on foreign chip supply chains. SoftBank’s ownership of key semiconductor companies gives Japan an influential position in that emerging landscape.
Graphcore’s growing presence in India also fits into that broader strategy. Last year, the company announced plans to invest up to £1 billion in a new AI campus in Bengaluru, hiring hundreds of engineers across silicon, systems, and software development.
India is rapidly becoming a crucial battleground for AI infrastructure because of its large engineering workforce, expanding digital economy, and government push into semiconductor manufacturing.
However, the challenge now for Graphcore is proving that it can regain relevance in a market increasingly dominated by a handful of giant players. Nvidia still commands overwhelming influence in AI training and inference chips, while rivals, including Advanced Micro Devices, Intel, and several startups, are all competing aggressively for market share.
But SoftBank’s backing has given Graphcore something many smaller AI hardware firms lack: deep-pocketed financial support and integration into a wider strategic ecosystem.
The broader question, however, is whether even billions of dollars and strong technology are enough to break Nvidia’s dominance. The AI boom has shown that success in semiconductors depends not only on chip performance but also on software ecosystems, developer adoption, manufacturing scale, and long-term customer relationships.