SoftBank has poured $450 million into UK-based AI chip company Graphcore, a move that could reshape the semiconductor market and help ease the GPU shortage that has slowed AI development worldwide. The investment positions Graphcore as a stronger rival to Nvidia, whose chips have dominated the AI training and inference space.
Why Graphcore matters in the AI chip race
Graphcore, based in Bristol, designs processors specifically for AI workloads. Its Intelligence Processing Unit (IPU) differs from traditional GPUs, which were originally built for graphics. The company has long been seen as one of the few credible challengers to Nvidia, but it has struggled to gain traction among major cloud providers. SoftBank's cash injection gives Graphcore the runway to scale its manufacturing and win new customers.
What the investment means for GPU supply
AI companies have faced months-long wait times for Nvidia's H100 and Blackwell chips. Graphcore's IPUs, if they can ramp up production, could offer an alternative for tasks like natural language processing and computer vision. The investment may help ease those shortages by adding more AI-specialized silicon to the market. SoftBank's deep pockets also mean Graphcore can afford to build the inventory that hyperscalers demand.
Challenging Nvidia's grip
Nvidia controls roughly 80% of the AI accelerator market, according to analyst estimates. SoftBank's bet on Graphcore signals a belief that the market can support a second large-scale player. The Japanese conglomerate has a history of making bold chip bets—it paid $31.6 billion for Arm in 2016. This time, it's targeting a smaller, more focused company that could carve out a niche in a sector where Nvidia has grown increasingly dominant.
The deal is expected to close by the end of the year. Graphcore has said it will use the funds to accelerate production of its next-generation chip, the GC200, and expand its engineering team. Whether that will be enough to dent Nvidia's lead remains an open question.
