AMD unveils AI server in challenge to Nvidia's dominance as OpenAI taps its newest chips
SAN JOSE - Advanced Micro Devices chief executive officer Lisa Su on June 12 unveiled a new artificial intelligence server for 2026 that aims to challenge Nvidia's flagship offerings as OpenAI's CEO said the ChatGPT creator would adopt AMD's latest chips.
Ms Su took the stage at a developer conference in San Jose, California, called 'Advancing AI' to discuss the MI350 series and MI400 series AI chips that she said would compete with Nvidia's Blackwell line of processors
The MI400 series of chips will be the basis of a new server called 'Helios' that AMD plans to release in 2026.
The move comes as the competition between Nvidia and other AI chip firms has shifted away from selling individual chips to selling servers packed with scores or even hundreds of processors, woven together with networking chips from the same company.
The AMD Helios servers will have 72 of AMD's MI400 series chips, making them comparable to Nvidia's current NVL72 servers, AMD executives said.
During its keynote presentation, AMD said that many aspects of the Helios servers - such as the networking standards - would be made openly available and shared with competitors such as Intel. The move was a direct swipe at market leader Nvidia, which uses proprietary technology called NVLink to string together its chips but has recently started to license that technology as pressure mounts from rivals.
'The future of AI is not going to be built by any one company or in a closed ecosystem. It's going to be shaped by open collaboration across the industry,' Ms Su said.
Ms Su was joined onstage by OpenAI's Sam Altman. The ChatGPT creator is working with AMD on the firm's MI450 chips to improve their design for AI work.
'Our infrastructure ramp-up over the last year, and what we're looking at over the next year, have just been a crazy, crazy thing to watch,' Mr Altman said.
During her speech, executives from Elon Musk-owned xAI, Meta Platforms and Oracle took to the stage to discuss their respective uses of AMD processors. Crusoe, a cloud provider that specializes in AI, told Reuters it is planning to buy US$400 million (S$511 million) of AMD's new chips.
AMD's Ms Su reiterated the company's product plans for 2026, which will roughly match the annual release schedule that Nvidia began with its Blackwell chips.
AMD shares ended 2.2 per cent lower after the company's announcement. Kinngai Chan, an analyst at Summit Insights, said the chips announced on June 12 were not likely to immediately change AMD's competitive position.
AMD has struggled to siphon off a portion of the quickly growing market for AI chips from the dominant Nvidia. Its software called ROCm has struggled to gain traction against Nvidia's CUDA, which is seen by some industry insiders as a key part of protecting the company's dominance.
But AMD has made a concerted effort to improve its software and produce a line of chips that rival Nvidia's performance. AMD completed the acquisition of server builder ZT Systems in March. As a result, AMD is expected to launch new complete AI systems, similar to several of the server-rack-sized products Nvidia produces.
AMD has also made a series of small acquisitions in recent weeks and has added talent to its chip design and AI software teams. At the event, Ms Su said the company has made 25 strategic investments in the past year that were related to the company's AI plans.
When AMD reported earnings in May, Ms Su said that despite increasingly aggressive curbs on AI chip exports to China, AMD still expected strong double-digit growth from AI chips. REUTERS
Join ST's Telegram channel and get the latest breaking news delivered to you.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
CNA
3 hours ago
- CNA
CNA Explains: How Nvidia grew to become the world's No 1 company
Scroll up for the next video X CNA Explains: How Nvidia grew to become the world's No 1 company
CNA
7 hours ago
- CNA
Foxconn's Apple era fades as AI servers drive growth in Taiwan tech sector
TAIPEI :Taiwan's Foxconn, which rose to become a global tech manufacturing juggernaut by assembling millions of iPhones, can now say its main business is no longer Apple as it takes advantage of the AI-boom to diversify its income. Its revenue from making AI servers and other cloud and networking products, including for major customer Nvidia, surpassed smart consumer products such as iPhones for the first time in the second quarter, marking the culmination of a shift that began years ago and has swept through Taiwan's tech industry. Foxconn's heavy reliance on the smartphone business has long been viewed by investors as a significant risk, as demand growth for new iPhones has gradually weakened since they were first introduced nearly two decades ago, leaving the top iPhone assembler grappling with slowing sales momentum, analysts said. Wary of the risk, Foxconn Chairman Young Liu has been championing new businesses such as AI servers, electric vehicles and semiconductors since taking the top job in 2019. While its expansion into EVs and chips has yet to show a meaningful contribution to its topline, Foxconn's success in AI server manufacturing - the company is Nvidia's biggest server maker - is the result of its early bets before the technology was thrust into the limelight with the advent of ChatGPT in late 2022. Consumer electronics accounted for 35 per cent of Foxconn's total revenue in the second quarter, while cloud and networking business represented 41 per cent. In 2021, consumer electronics represented 54 per cent of its revenue. The firm's prudent wagers years back helped it cultivate a now-prized relationship with the U.S. AI chip firm and other major AI players, analysts said. "The company has been in the business for years, meeting higher quality requirements, diversifying assembly and operations across sites, and pursuing vertical integration,' said Ming-Chi Kuo, an analyst at TF International Securities. Foxconn began producing reference designs for Nvidia's graphics cards around 2002 and started making general-purpose servers for cloud service providers' data centres as early as around 2009. Its AI server business with Nvidia is in many ways the culmination of that history, analysts said. Foxconn says it is now one of the world's largest suppliers of both general-purpose and AI servers, with a market share of nearly 40 per cent in each. The company has also shown a willingness to commit investment to a project at an earlier stage than other companies, Kuo said, citing its past investments for Apple and similar moves for Nvidia. 'In long-term partnerships, Foxconn is more willing to take the initiative,' he said. Foxconn's plan to build factories in Houston, Texas — part of Nvidia's $500 billion U.S. investment plan — and in Mexico to produce AI servers for the U.S. client underscores this strategy, analysts said. Foxconn now expects its AI server revenue would grow more than 170 per cent in the third quarter year-on-year. Foxconn and Nvidia declined to comment. Apple did not respond to request for comment. BROADER SHIFT The shift at Foxconn mirrors a broader trend in Taiwan's technology sector, where companies once centred on consumer electronics — such as Foxconn with iPhones, and Quanta Computer and Wistron Corp with notebooks — are now investing heavily in AI servers. Nvidia partner Wistron's revenue for January to July rose 92.7 per cent, while Quanta's grew 65.6 per cent in the same period. "The monthly sales jump for Taiwan ODMs in the first half of 2025 is evidence of this trend,' said Robert Cheng, head of Asia technology hardware research at BofA Global Research, referring to original design manufacturers like Foxconn that contract manufacture products for their clients. Their fast transition into AI servers is also the result of Taiwanese tech supply chain working closely with U.S. tech giants on data centre infrastructure work for a decade now, according to Chris Wei, industry consultant at Taiwan's Market Intelligence & Consulting Institute. He estimates Taiwan accounts for about 80 per cent of global server shipments and more than 90 per cent of AI servers. Cheng agrees.
CNA
9 hours ago
- CNA
Foxconn and SoftBank to make data centre equipment in Ohio for Stargate project
TAIPEI :Foxconn's said on Monday it plans to manufacture data centre equipment with Japan's SoftBank at the Taiwanese firm's former electric vehicle factory in Ohio, part of the Stargate project to advance U.S. artificial intelligence infrastructure. Softbank has purchased the Lordstown site, Foxconn Chairman Young Liu told reporters, adding that Foxconn will continue operating the site via a venture the two companies will set up. Foxconn said earlier this month it had struck a deal to sell the factory and its machinery for $375 million, without naming the buyer. Stargate is a joint venture between SoftBank, ChatGPT's creator OpenAI and Oracle. The project was announced by U.S. President Donald Trump in January, who said that the companies would invest up to $500 billion. SoftBank did not immediately respond to a request for comment. SoftBank and Foxconn started preparatory work for the project more than half a year ago, Young added. "We understand that for this project, the first priorities are power, venue, and timing — it cannot be delayed for too long. Taking all these factors into account, we believe Ohio is a very suitable location, and SoftBank shares this view," he said.
