And there was way more introduced on the present, together with new InfiniBand networking playing cards and switches, important upgrades and new options in its Enterprise AI software program, and a storage companion validation service.
Nvidia additionally launched its next-generation AI supercomputer, the DGX SuperPOD10. Powered by GB200 Grace Blackwell superchips, the SuperPOD10 is able to processing trillion-parameter fashions for large-scale generative AI with 11.5 exaFLOPS utilizing FP4 knowledge varieties and 240 TB of quick reminiscence.
Nvidia inventory splits
For its first fiscal quarter 2025, which ended on April 28, 2024, Nvidia reported income of $26 billion, up 18% from the earlier quarter and up 262% year-over-year. Working revenue surged to $16.9 billion, a 24% improve from the earlier quarter and a 690% improve from the yr prior. GAAP earnings per diluted share have been $5.98, a 21% improve from the earlier quarter and a 629% improve from a yr in the past. All of those numbers considerably exceeded expectations from Wall Road analysts.
Nvidia additionally introduced a 10-to-1 inventory break up, efficient in June. Nvidia inventory had seen a large run as much as over $1,000 per share.
Blackwell techniques emerge, Rubin is teased
On the huge Computex IT {hardware} present, Nvidia unveiled new Blackwell-powered techniques that, it stated, will enable enterprises to construct “AI factories” and knowledge facilities to drive the following wave of generative AI.
Nvidia additionally introduced that its Nvidia MGX modular reference design platform, introduced on the 2003 Computex present, now helps Blackwell merchandise. It launched the brand new Nvidia GB200 NVL2 platform, a smaller model of the GB200 NVL72 launched in March, that it stated will pace up knowledge processing by as much as 18x, with 8x higher vitality effectivity in comparison with utilizing x86 CPUs.
