Nvidia’s Chief Executive Officer Jensen Huang has outlined ambitious plans to put AI at the centre of the company’s technology development strategy.
At a keynote address at National Taiwan University, ahead of the Computex 2024 event, Nvidia announced a Blackwell Ultra chip for 2025 and a next-generation platform in development called Rubin for 2026.
The upcoming Rubin AI platform will use HBM4, the next iteration of the essential high-bandwidth memory. The Rubin family of chips will include new graphics and central processors as well as networking chips.
Nvidia recently announced plans to begin shipping its new Blackwell GPU system, but used the event on Sunday to add that it would update that GPU to Blackwell Ultra by next year.
He cited generative AI as a new industrial revolution and said that he expects it to play a major role as the technology shifts to focusing on personal computers. Already, Microsoft recently announced its AI plans that focus heavily on generative AI with several PC manufacturers announcing a slew of new products that have generative AI capabilities mainly backed by Qualcomm chips.
Nvidia is best known for its artificial intelligence data centre systems and is now looking to broaden its customer base beyond the handful of cloud-computing giants and enlist new customers.
He said Nvidia’s chip solutions for AI factories will unlock a A$150.27 trillion in opportunities across computing, transportation, healthcare and manufacturing.
“We are seeing computation inflation,” said Huang. He noted that as the amount of data that needs to be processed grows exponentially, traditional computing methods cannot keep up and it’s only through Nvidia’s accelerated computing costs can be cut down.
He touted 98 per cent cost savings and 97 per cent less energy required with Nvidia’s technology, saying that constituted “CEO math, which is not accurate, but it is correct,” noted Bloomberg.
Huang noted that the company now plans to release a new family of AI chips every year, accelerating its prior release schedule of roughly every two years.
“Our company is on a one-year rhythm,” Nikkei Asia reported Huang as having said. He added that Nvidia intends to “push everything” to its limits in terms of chipmaking and will “sell to [customers] on a one-year rhythm.”
Huang announced collaboration with key Taiwanese electronics suppliers including Foxconn, Pegatron, Wistron and Delta Electronics to adopt Nvidia’s AI technologies.
During Sunday’s speech, Huang pointedly noted Taiwan’s central role in the semiconductor manufacturing industry and its importance to global supply chains, with the island’s production playing a crucial role in the development of Nvidia’s AI computing chips.
