A Dell executive has teased a blockbuster GPU to be launched by Nvidia later this year.
Nvidia, the leading global supplier of chips used in artificial intelligence (AI), is getting ready to unleash the B100 which will blow the H100 out of the water.
That impressive commodity, running at 700W, will be outstripped by its successor with the B100 set to consume 1000 watts – an increase of 42%.
Nvidia is said to enjoy an enviable 80% market share in the high-end AI chip sector, but it plans further expansion including the formation of a dedicated unit for designing custom chips for cloud computing companies and other sectors.
Jeff Clarke, COO at Dell Technologies, has expressed his anticipation for the upcoming AI accelerator, speaking on his company’s recent earnings call.
“We’re excited about what’s happening with the H200 and its performance improvement,” he told investors, with the same sentiment expressed toward the B100 chip and an unknown B200.
When will Nvidia launch the B100 chip?
It was claimed direct cooling liquid won’t be required for GPU’s sapping through 1000 watts, but it could be introduced “next year with the B200”.
This aspect of the dialogue has provoked curiosity and intrigue as there is no current chip with that label on the horizon at Nvidia, at least not on the roadmap shared with its investors in October 2023.
It could be Clarke was referring to the GB200 Superchip which is expected to bring together Nvidia’s Grace CPU with its B100 GPU, like the GH200. Alternatively, a trump card could be up the proverbial sleeve of the Delaware-incorporated tech multinational.
The company declined to comment, following the remarks made by the Dell chief, but further information will be forthcoming at Nvidia’s GTC event from 17-21 March.
The arrival of the B100 isn’t expected until late this year, given the launch of to launch of the H200 GPU in the first half of the year.
Image: Marketing Maverick/X
