AMD Stakes Future on Open AI Infrastructure
Top-tier AI customers including OpenAI, Meta, Microsoft, Oracle, xAI and Crusoe pledged significant investments. OpenAI's CEO Sam Altman joined Su onstage, confirming the firm's shift to MI400-class chips and collaboration on MI450 design. Crusoe disclosed a $400 million commitment to the platform.
MI350 Series, which includes the MI350X and MI355X, are shipping to hyperscalers now, with a sharp generational performance leap — delivering about four times the compute capacity of prior-generation chips, paired with 288 GB of HBM3e memory and up to 40% better token‑per‑dollar performance than Nvidia's B200 models. Initial deployments are expected in Q3 2025 in both air‑ and liquid‑cooled configurations, with racks supporting up to 128 GPUs, producing some 2.6 exaflops FP4 compute.
ADVERTISEMENT
Looking further ahead, AMD previewed 'Helios'—a fully integrated rack comprising MI400 GPUs, Zen 6‑based EPYC 'Venice' CPUs and Pensando Vulcano NICs, boasting 72 GPUs per rack, up to 50% more HBM memory bandwidth and system‑scale networking improvements compared to current architectures. Helios is poised for market launch in 2026, with an even more advanced MI500‑based variant expected around 2027.
Dr Su underscored openness as AMD's competitive lever. Unlike Nvidia's proprietary NVLink interface, AMD's designs will adhere to open industry standards—extending availability of networking architectures to rivals such as Intel. Su argued this approach would accelerate innovation, citing historical parallels from open Linux and Android ecosystems.
On the software front, the ROCm 7 stack is being upgraded with enterprise AI and MLOps features, including integrated tools from VMware, Red Hat, Canonical and others. ROCm Enterprise AI, launching in Q3 or early Q4, aims to match or exceed Nvidia's CUDA-based offerings in usability and integration.
Strategic acquisitions underpin AMD's infrastructure ambitions. The purchase of ZT Systems in March 2025 brought over 1,000 engineers to accelerate rack-scale system builds. Meanwhile, AMD has onboarded engineering talent from Untether AI and Lamini to enrich its AI software capabilities.
Market reaction was muted; AMD shares fell roughly 1–2% on the event day, with analysts noting that while the announcements are ambitious, immediate market share gains are uncertain.
Financially, AMD projects AI data centre revenues growing from over $5 billion in 2024 to tens of billions annually, anticipating the AI chip market reaching around $500 billion by 2028.
These developments position AMD as a serious contender in the AI infrastructure arena. Its push for rack‑scale systems and open‑standard platforms aligns with the growing trend toward modular, interoperable computing. Competition with Nvidia will intensify through 2026 and 2027, centred on performance per dollar in large‑scale deployments.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Khaleej Times
12 hours ago
- Khaleej Times
SoftBank's growing bets on AI and semiconductor assets
SoftBank announced on Tuesday a $2 billion investment in Intel that will add to the Japanese tech investor's ambitious bet on artificial intelligence and the chip industry. Below are major AI investments SoftBank has announced so far: INTEL On August 18, SoftBank announced a $2 billion equity investment in Intel, saying the deal reflected its belief that chip manufacturing and supply would expand further in the U.S., and the U.S. company would play a critical role. STARGATE Stargate is a joint venture between SoftBank, ChatGPT's creator OpenAI, and Oracle. U.S. President Donald Trump announced the project in January, saying that the companies would invest up to $500 billion. On August 7, SoftBank CFO Yoshimitsu Goto said the project, for which SoftBank is leading financing, had been delayed by longer than expected negotiations with other parties and decisions on sites. Goto said at the time SoftBank remained committed to its target of building $500 billion worth of data centres over four years and the banks involved in the project financing, including Japan's "megabanks" and major U.S. lenders, had shown positive interest in the scheme. SoftBank has yet to release details on what kinds of returns its financing of the Stargate project could generate. The extent of third-party investment will determine what other financing tools, such as bank loans and debt issuance, it may have to deploy. On August 18, Taiwan's Foxconn said that it planned to manufacture data centre equipment with SoftBank at the Taiwanese firm's former electric vehicle factory in Ohio as part of the Stargate project. OPENAI SoftBank is leading a $40 billion funding round for OpenAI. It has until the end of 2025 to fund its $22.5 billion portion, although the remainder has been subscribed, according to a source familiar with the matter. PERPLEXITY AI The three-year-old startup that operates an AI-powered search engine has raised about $1 billion in funding so far from investors including Nvidia and SoftBank. It was last valued at $14 billion. AMPERE COMPUTING SoftBank announced a $6.5 billion deal in March to acquire Ampere Computing, a U.S. chip designer startup founded by the former president of Intel. Ampere makes data centre central processing unit chips based on computing architecture from SoftBank's majority-owned Arm Holdings that are used by firms such as Oracle in their cloud computing infrastructure. GRAPHCORE In 2024, SoftBank bought the British AI chipmaker for an undisclosed sum. Once touted as a rival to Nvidia, Graphcore had struggled to secure the investment needed to compete. ARM HOLDINGS The British chip designer owned by SoftBank went public in 2023, securing a valuation of $54.5 billion. SoftBank still owns the majority of the company. Arm-based devices power nearly every smartphone in the world and server chips based on its intellectual property have made significant inroads in the data centre market. As part of a broad plan to increase its business, Arm has sought to expand beyond supplying crucial chip intellectual property to building its own complete designs. SoftBank had owned around 5% of Nvidia until it sold the stake in 2019, before ChatGPT generated a surge in AI interest at the end of 2022. It has been building up its stake again and held Nvidia shares worth around $4.8 billion as of the end of June this year, according to an SEC filing.
Tahawul Tech
15 hours ago
- Tahawul Tech
AI player Anthropic teams up with U.S. Government branches
Anthropic, an AI start-up, has joined forces with the US General Services Administration (GSA) to offer its Claude AI model to all branches of the federal government for $1. This strategy echoes a similar move by competitor OpenAI earlier in the year. The AI player will offer Claude for Enterprise and Claude for Government to federal civilian executive, legislative and judiciary branches. Additionally, agencies will have access to Anthropic's frontier models with continuous updates as new capabilities are released. It will provide technical support to agencies to integrate AI into their workflows quickly to promote modernisation and improved government productivity. On 5 August, GSA, the government's centralised procurement division, announced it added Anthropic, Google and OpenAI to its list of approved AI companies for use across civilian federal agencies. 'By offering expanded Claude access across all three branches of government, we're helping the federal workforce leverage frontier AI capabilities to maintain our competitive advantage and better serve the American people,' stated Anthropic CEO Dario Amodei. OpenAI's ChatGPT, Google's Gemini and xAI's Grok are also taking part in GSA's Multiple Award Schedule (MAS) contracting platform for use across civilian federal agencies. The models by the AI heavyweights come with contract terms in place, which allows the federal agencies to adopt and use them at a rapid pace. While the companies benefit from having their AI models widely available to the federal agencies, GSA stated the partnerships support US President Donald Trump's AI action plan, designed to solidify the US' position as the global leader in the face of competition from countries such as China. Source: Mobile World Live Image Credit: Anthropic
Zawya
18 hours ago
- Zawya
OpenAI rolls out cheapest ChatGPT plan at $4.6 in India to chase growth
ChatGPT maker OpenAI on Tuesday launched ChatGPT Go, a new India-only subscription plan priced at 399 rupees ($4.57) per month, its most affordable offering yet, as the company looks to deepen its presence in its second-largest market. Global companies often offer cheaper subscription plans for India's price-sensitive market, targeting the nearly one billion internet users in the world's most populous nation. The plan allows users to send up to ten times more messages and generate ten times more images compared to the free version, while also offering faster response times. Message limits increase with higher-tier subscription plans. ChatGPT Go is designed for Indians who want greater access to ChatGPT's advanced capabilities at a more affordable price, the Microsoft-backed startup said in a statement. The top-tier version of ChatGPT - ChatGPT Pro - is priced at 19,900 rupees/month in India, while ChatGPT Plus, its mid-range plan, costs 1,999 rupees/month. Earlier this year, CEO Sam Altman met with India's IT minister and discussed a plan to create a low-cost AI ecosystem. India is OpenAI's second-largest market by user base after the United States and may soon become the biggest, Altman said recently. ($1 = 87.3520 Indian rupees) (Reporting by Kashish Tandon in Bengaluru; Editing by Sonia Cheema)
