- Nvidia CEO confirms Blackwell Extremely and Vera Rubin names
- Each set to be revealed at GTC 2025 in March 2025
- Nvidia sees general revenues develop 114% year-on-year as AI demand booms
Nvidia CEO Jensen Huang has teased that the corporate’s subsequent flagship chip can be proven off inside the subsequent few weeks – in addition to apparently formally confirming its identify.
Talking to analysts on the corporate’s fiscal This fall 2025 earnings name, Huang revealed the brand new {hardware} will bear the moniker of Blackwell Extremely, and can be unveiled at its Nvidia GTC 2025 occasion in March 2025.
“Come to GTC and I’ll discuss to you about Blackwell Extremely, Vera Rubin, after which present you the one click on after that,” Huang declared.
Nvidia Blackwell Extremely…and extra
Huang went on so as to add Blackwell Extremely can be formally launched within the second half of 2025, and provide upgrades in processors, networking, and reminiscence – however can be constructed on the identical system structure as Blackwell.
Each Blackwell Extremely and Vera Rubin had been teased on an organization roadmap doc at Computex 2024, however Huang has now seemingly confirmed the naming of each.
Nvidia unveiled Blackwell at GTC 2024, promising a serious step ahead when it comes to AI energy and effectivity.
It has since seen various releases, together with the primary Blackwell “superchip”, the GB200, which has the power to scale up from a single rack all the way in which to a complete knowledge heart, as Nvidia appears to be like to push on with its management within the AI race.
Blackwell accommodates 208 billion transistors (up from 80 billion in Hopper) throughout its two GPU dies, that are related by 10 TB/second chip-to-chip hyperlink right into a single, unified GPU, making it as much as 30x quicker than Hopper in terms of AI inference duties, providing as much as 20 petaflops of FP4 energy, far forward of anything available on the market right this moment.
Regardless of this, Nvidia says Blackwell can scale back value and vitality consumption by as much as 25x, giving the instance of coaching a 1.8 trillion parameter mannequin – which might beforehand have taken 8,000 Hopper GPUs and 15 megawatts of energy – however can now be achieved by simply 2,000 Blackwell GPUs consuming simply 4 megawatts.
Vera Rubin is the *subsequent* step ahead for Nvidia after Blackwell Extremely, with an anticipated launch in 2026, with CPU and GPU merchandise anticpated, together with a Vera Rubin board combining the GPU and CPU right into a “superchip”.
Huang’s tease got here as Nvidia revealed its most up-to-date set of economic outcomes, with the corporate as soon as once more seeing file returns because it advantages from the large surge in AI demand.
Total Nvidia revenues greater than doubled year-on-year, reaching $130.5 billion for the monetary 12 months, a 114% rise, largely helped by file knowledge heart revenues of $35.6 billion, up 16% from the earlier quarter, and up 93% from a 12 months in the past.
“Demand for Blackwell is wonderful as reasoning AI provides one other scaling legislation — growing compute for coaching makes fashions smarter and growing compute for lengthy pondering makes the reply smarter,” Huang stated of the outcomes.
“We’ve efficiently ramped up the massive-scale manufacturing of Blackwell AI supercomputers, attaining billions of {dollars} in gross sales in its first quarter. AI is advancing at mild pace as agentic AI and bodily AI set the stage for the following wave of AI to revolutionize the biggest industries.”
By way of The Verge