Topics
Latest
AI
Amazon
Image Credits:David Paul Morris / Bloomberg / Getty Images
Apps
Biotech & Health
mood
Image Credits:David Paul Morris / Bloomberg / Getty Images
Cloud Computing
Commerce
Crypto
Enterprise
EVs
Fintech
Fundraising
contraption
game
Government & Policy
Hardware
Layoffs
Media & Entertainment
Meta
Microsoft
privateness
Robotics
Security
societal
Space
inauguration
TikTok
Transportation
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
Podcasts
video
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
Onstage at Nvidia ’s GTC 2025 conference in San Joseon Tuesday , CEO Jensen Huang foretell a hatful of new GPUs coming down the company ’s product pipeline over the next few month .
Perhaps the most significant is Vera Rubin . Vera Rubin , which is set to be released in the second half of 2026 , will feature tens of gigabytes of memory board and a customs Nvidia - design CPU yell Vera . Vera Rubin delivers substantial performance uplift compare to its forerunner , Grace Blackwell , Nvidia exact , particularly on AI inferencing and training undertaking .
When paired with Vera , Rubin — which is two GPUs in one , technically — can make out up to 50 petaflops while doing inference ( i.e. , running AI models ) , more than double the 20 petaflops for Nvidia ’s current Blackwell chip . Moreover , Vera is about twice as fast as the processor used in Nvidia ’s Grace Blackwell GPU .
Rubin will be stick to by Rubin Ultra in the second half of 2027 , a collection of four GPUs in a single package delivering up to 100 petaflops of performance .
On the close horizon — H2 2025 — Nvidia will liberate Blackwell Ultra , a GPU that ’ll fall in several configurations . A single Ultra chip will bid the same 20 petaflops of AI performance as Blackwell , but with 288 GB of memory — up from 192 GB in vanilla extract Blackwell .
On the far horizon are Feynman GPUs . Huang during the tonic gave few item about Feynman ’s architecture , key after American theoretical physicist Richard Feynman — write that it boast a Vera CPU . Nvidia design to impart Feynman , which will succeed Rubin Vera , to market sometime in 2028 .
Updated 3/18 3:07 p.m. Pacific : An former version of this story indicated that Vera Rubin had “ tens of terabytes ” of computer memory . In fact , it has “ tens of G ” of memory . We regret the error .
