Nvidia’s improved GB300 GPU is powered by the B300 chip and will be available March 2025. It will give a 1.5x increase in FP4 performance. Its ultra-architecture also facilitates a substantial memory capacity upgrade from 192GB HBM2e (8 layers) to 288GB HBM3e (12 layers), enabling faster processing of massive datasets and improving overall performance.
It will use 200W additional power with TDP going to 1.4KW and 1.2KW for the GB300 and B300 HGX respectively (compared to 1.2KW and 1KW for GB200 and B200).
More memory for H200 chips showed a 3X improvement in cost for AI reasoning models vs H100. The extra memory for GB300 should have a larger benefit for AI reasoning performance vs B200 chips.
Nvidia NVL72 enables 72 GPUs to work on the same problem, sharing their memory, at extremely low latency. This enables lower cost with larger batches and more intelligence with longer chain of thought reasoning.
The tokenomics with NVL72 are more than 10x better.
Microsoft as one of the slowest to deploy GB300 due to design speed, with them still buying some GB200 in Q4.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.