AMD Unveils New AI Chips to Compete with Nvidia

AMD has unveiled its latest advancements in AI and server technologies during a high-profile event in San Francisco. The company introduced the MI325X AI chip, set to enter mass production in the fourth quarter of 2024, and the MI350 series chips, which are planned for launch in the second half of 2025 with enhanced performance capabilities. These new AI chips are part of AMD's strategy to expand its presence in a market dominated by Nvidia, responding to the surging demand for AI processors among major tech firms like Microsoft and Meta.

MI325X and MI350 Series Chips

The new MI325X AI chip will start mass production in Q4 2024 and is based on the same architecture as the MI300X, which was launched last year. This updated chip features a new type of memory, which AMD claims will significantly speed up AI calculations. The MI350 series, launching in the second half of 2025, will include increased memory and a new underlying architecture, promising even better performance compared to the previous MI300X and MI250X models. These advancements aim to compete directly with Nvidia's Blackwell architecture, providing more options for AI processing to meet growing market demands.

Expanding Global Availability

Santa Clara-based AMD has also announced that vendors like Super Micro Computer will begin shipping the new MI325X AI chips to customers in early 2025. This marks an important step in AMD's efforts to increase the availability of its AI products worldwide. The company is actively positioning itself as a strong competitor to Nvidia by enhancing the capabilities of its AI chips and addressing the ongoing supply shortages faced by tech giants eager to implement AI solutions.

Advancements in Server CPUs

In addition to AI chips, AMD unveiled a new version of its server CPU family, formerly codenamed Turin. The flagship chip features nearly 200 processing cores and uses the Zen 5 architecture, which promises up to 37% speed gains for AI data processing. One of the new server chips is specifically designed to feed data to graphics processing units (GPUs), thereby enhancing AI processing efficiency. The flagship CPU is priced at $14,813, reflecting its advanced capabilities and potential impact on AI-driven workloads.

Market Reaction and AI Chip Forecast

Despite the announcements, AMD shares fell 2.5% during the event, while Nvidia’s shares gained 1.2%. Analysts note that AMD's latest chip launch is unlikely to immediately impact Nvidia's dominance in data center revenue, given the substantial demand for AI chips. In July, AMD raised its AI chip revenue forecast for the year from $4 billion to $4.5 billion, driven by increased interest in the MI300X chips amid the ongoing AI boom.

Wall Street estimates suggest that AMD's data center revenue will reach $12.83 billion this year, while Nvidia is expected to report data center revenue of $110.36 billion. This discrepancy underscores the scale of Nvidia's lead in the AI chip market, although AMD is making strides to carve out a larger share with its expanding AI product offerings.

The Future of AI Processing

AMD's new AI and server chips are a significant step forward in the company's efforts to compete with Nvidia and capture a share of the growing AI market. With new advancements in chip memory, architecture, and server integration, AMD aims to provide competitive alternatives for enterprises looking to implement AI technologies. As the demand for AI processors continues to surge, AMD's focus on enhancing both AI and server technologies could help bridge the gap in availability and offer customers more choices in a rapidly evolving market.