Nvidia unveiled what it claimed is the world’s most powerful chip for AI, one of a slew of announcements from the opening day of its annual developer conference which also included a 6G research platform and a set of enterprise software tools.

Company chief Jensen Huang (pictured) told a keynote audience Nvidia’s flagship B200 Blackwell chip can perform certain tasks 30-times faster than its predecessor and is designed to increase the computing power for large language models.

B200 Blackwell includes 208 billion transistors compared with 80 billion in the Hopper H100 chip unveiled in 2023.

The Hopper platform helped Nvidia become the runaway leader in AI chips, with the company currently holding an 80 per cent market share.

Compared with current GPUs, Huang explained the Blackwell chip was twice as powerful in training AI models, 30-times speedier in performing certain tasks like responding to chatbots, and reduced costs and energy consumption by 25 times.

Huang said “generative AI is the defining technology of our time” and Blackwell would prove to be engine to “power this industrial revolution”.

Quote Icon

Working with the most dynamic companies in the world, we will realise the promise of AI for every industry.

Jensen Huang – CEO Nvidia

Companies including Meta Platforms, Google, Microsoft, OpenAI and Tesla are expected to adopt Blackwell, Nvidia stated.

6G
The US tech giant also took the wraps off a platform designed to assist researchers in using AI for future 6G applications, with tools tailored to work around digital twinning and network testing.

Nvidia stated its 6G Research Cloud Platform offers foundational tools to boost the use of AI in developing RAN technology, helping organisations with their 6G research.

Nokia, SoftBank Corp, Fujitsu and Viavi Solutions are among the first adopters.

Nvidia also highlighted microservices, software tools tailored to make it easier for enterprises to use and deploy AI models in their own platforms.

Other announcements made at the event include a set of chips for the automotive sector capable of running chatbots in vehicles and a range focused on humanoid robots.