Nvidia unveils new flagship chip for AI, the B200


by Max A. Cherney and Stephen Nellis

SAN JOSE, Calif. (Reuters) – Nvidia, the artificial intelligence (AI) semiconductor giant, unveiled its new flagship product, the Blackwell B200 chip, on Monday on the first day of its annual developer conference (GTC). ).

This new chip and the various announcements expected at this conference will give indications on Nvidia’s ability to maintain its lead in the field of components for AI, the growth of which has fueled the rise in the New York Stock Exchange in recent months. York. Nvidia dominates the market for artificial intelligence chips for data centers, with a market share of around 80% last year.

The B200 was introduced by the group’s general manager, Jensen Huang, who arrived on stage wearing his usual black leather jacket.

Read alsoCounting

It is equipped with 208 billion transistors, more than 2.5 times the 80 billion of the group’s previous chip. All of these transistors can access the memory at almost the same time, which increases productivity.

“The launch of Blackwell means that Nvidia will capture even more capital spending dollars than with its previous generation Hopper. We expect Nvidia to continue not only to dominate but to widen the gap with its competitors in the ‘AI,” Kinngai Chan, executive director and senior analyst at Summit Insights, said after the presentation.

Nvidia said big tech names such as Amazon, Alphabet subsidiary Google, Meta Platforms, Microsoft, OpenAI and Tesla are expected to use its new chip.

Jensen Huang also announced partnerships with design software companies Ansys, Cadence and Snopsys, whose shares immediately rose about 3% in after-hours trading on Wall Street.

A new darling of investors, Nvidia has seen its stock price soar by 240% over the last 12 months, becoming the third largest market capitalization in the United States behind Microsoft and Apple.

Its market share is nevertheless expected to decline by several points in 2024 with the arrival of new products developed by competitors such as Intel and Advanced Micro Devices.

It will now also have to respond to the immense expectations of investors for the development of its AI activity.

While the group is primarily known as a chip designer, it has also developed a whole series of software products.

Nvidia, which is based in Santa Clara, California, is holding its GTC conference in person this year for the first time since 2019 and expects to receive 16,000 attendees, about double the last in-person event.

(Reporting Max A. Cherney and Stephen Nellis in San Francisco; French version Alban Kacher, edited by Blandine Hénault and Bertrand Boucey)

©2024 Thomson Reuters, all rights reserved. Reuters content is the intellectual property of Thomson Reuters or its third party content providers. Any copying, republication or redistribution of Reuters content, including by framing or similar means, is expressly prohibited without the prior written consent of Thomson Reuters. Thomson Reuters shall not be liable for any errors or delays in content, or for any actions taken in reliance thereon. “Reuters” and the Reuters Logo are trademarks of Thomson Reuters and its affiliated companies.



Source link -87