(Bloomberg) — Nvidia Corp. CEO Jensen Huang said the company plans to upgrade its AI accelerators every year, announcing a Blackwell Ultra chip for 2025 and a next-generation platform in development called Rubin for 2026.
Most read on Bloomberg
The company – now best known for its artificial intelligence data center systems – also showed off new software tools and models on the eve of the Computex show in Taiwan. Nvidia views the rise of generative AI as a new industrial revolution and hopes to play a major role in the technological transition to personal computers, the CEO said during a speech at National Taiwan University.
Nvidia has been the biggest beneficiary of a massive influx of AI spending, helping to make the company the world’s most valuable chipmaker. But it is now looking to expand its customer base beyond the handful of cloud computing giants that generate much of its sales. As part of this expansion, Huang expects more companies and government agencies to adopt AI – from shipbuilders to drug developers. He returned to themes he outlined a year ago at the same venue, including the idea that those without AI capabilities will be left behind.
“We are seeing an inflation of calculations,” Huang said on Sunday. As the amount of data to be processed increases exponentially, traditional computing methods cannot keep pace and only through Nvidia’s accelerated computing style can we reduce costs, Huang said. He touted 98% cost savings and 97% less power consumption from Nvidia’s technology, saying this constituted “CEO calculations, which are not accurate, but they are correct.”
Shares of Taiwan Semiconductor Manufacturing Co. and other suppliers rose after the announcement. TSMC shares rose as much as 3.9%, while Wistron Corp. gained 4%.
Huang said the next Rubin AI platform will use HBM4, the next iteration of the high-bandwidth mission-critical memory that has become a bottleneck for AI accelerator production, with leader SK Hynix Inc. largely sold out until 2025. He also did not offer detailed specifications. for upcoming products, which will follow Blackwell.
“I think introducing Rubin and Rubin Ultra was extremely smart and speaks to their commitment to a year-over-year refresh cycle,” said Dan Newman, CEO and chief analyst of Futurum Group. “What he emphasized most clearly, in my opinion, was the pace of innovation and the company’s relentless pursuit to maximize the limits of technology, including software, processes, packaging and partnerships, in order to protect and expand its moat and position in the market.
Nvidia has started selling gaming cards for desktop computers, and that context comes into play as computer makers work to add more AI features to their machines.
Microsoft Corp. and its hardware partners are using Computex to introduce new laptops with AI enhancements under the Copilot+ brand. The majority of these released devices are based on a new type of processor that will allow them to run longer on a single battery charge, provided by Nvidia’s rival Qualcomm Inc.
While these devices are suitable for simple AI features, adding an Nvidia graphics card will significantly increase their performance and bring new features to popular software like games, Nvidia said. PC makers such as Asustek Computer Inc. offer such computers, the company said.
To help software makers bring more new features to the PC, Nvidia offers pre-trained AI tools and models. They will handle complex tasks, like deciding whether to process data on the machine itself or send it to a data center over the Internet.
In addition, Nvidia is releasing a new design for server computers based on its chips. The MGX program is used by companies such as Hewlett Packard Enterprise Co. and Dell Technologies Inc. to help them more quickly bring products to market used by businesses and government agencies. Even competitors Advanced Micro Devices Inc. and Intel Corp. benefit from the design with servers that combine their processors with Nvidia chips.
Lisa Su, AMD’s CEO, took the stage at Computex the day after Huang’s comments, describing her company’s progress in AI chips. AMD is accelerating the introduction of its AI processors as it seeks to close the gap with Nvidia in this rapidly growing area.
Nvidia’s previously announced products, such as Spectrum It will also provide free access to NIM products. Microservices are a set of middleware software and models that help businesses deploy AI services faster, without having to worry about the underlying technology. Companies that deploy them must then pay Nvidia a usage fee.
Huang also encouraged the use of digital twins in a virtual world that Nvidia calls Ominverse. To show the scale possible, he showed a digital twin of planet Earth, called Earth 2, and how it can help achieve more sophisticated modeling of weather and other complex tasks. He noted that Taiwan-based contractors, such as Hon Hai Precision Industry Co., also known as Foxconn, use these tools to develop plans and operate their factories more efficiently.
–With help from Jane Lanhee Lee.
(Updates with analyst comment from sixth paragraph)
Most read from Bloomberg Businessweek
©2024 Bloomberg LP
