Softbank is shaping up to be a major force behind the chip designs of the next decade, if a report concerning its stake in Nvidia is correct.
A little less than a year after it bought mobile chip designer ARM for $35 billion, Softbank has slowly amassed a $4 billion stake in Nvidia, according to Bloomberg. The investment is part of a $100 billion fund that Softbank founder Masayoshi Son wants to invest in technology companies, and is just shy of the threshold that would require Softbank to disclose the stake.
Nvidia is currently considered the leading maker of chips for machine learning research, which is quite a path from the days when its primary goal was to render cool graphics in first-person shooters. All major cloud providers offer Nvidia GPUs to their customers for machine-learning applications, although last week Google signaled its intent to compete for those workloads with the reveal of its second-generation TPU (tensor processing unit) chip. Intel is also likely to compete in this arena, which means deep pockets will be needed to compete as more players target machine-learning workloads.
In what was probably a coincidence, Nvidia founder and CEO Jensen Huang posted an essay Wednesday morning detailing Nvidia’s approach to AI chip design, comparing and contrasting it with Google’s approach.