AMD to Mass Produce New AI Chip Later This Year

Advanced Micro Devices (AMD) announced on Thursday that it intends to begin mass producing a new version of its AI chip, the MI325X, in the fourth quarter of this year. This decision is a component of the company’s plan to increase and strengthen its position in market share in the AI chip industry, which is dominated by Nvidia.

At a presentation in San Francisco, Lisa Su, the CEO of AMD, a US semiconductor business, said that the company plans to introduce its next-generation MI350 family processors in the second half of 2025. These processors will include a new underlying architecture that AMD says will significantly improve performance over the previous MI300X and MI250X CPUs, in addition to having additional memory.

Based on AMD’s disclosures earlier this year, the revelations were mostly anticipated. They were unable to reassure investors, who caused AMD shares to drop by about 5% during afternoon trade. The lack of significant new cloud computing clients for the chips was cited by several experts as the reason for the decline. While Intel’s stock dipped 1.6%, competitor Nvidia’s stock increased 1.5%.

Major tech companies like Microsoft and Meta Platforms have been able to sell as many AI processors as they can since demand for these processors has been significantly higher than supply from Nvidia and AMD.

Over the past two years, this has caused a huge surge in chip equities; AMD’s shares have increased by almost 30% since a recent low in early August.

Kinngai Chan, a research analyst at Summit Insights, stated that “no new customers have been announced so far.” The stock had.

Super Micro Computer and other manufacturers will start distributing the MI325X AI processor to consumers in Q1 2025, AMD announced. The MI325X processor, which was introduced last year, has the same architecture as the MI300X chip, but it has new memory that speeds up AI processing.

To enhance data flow between chips and systems in data centres, the business has released a number of networking chips. A new iteration of the company’s server central processing unit (CPU) design was also introduced. The series of chips, formerly known as Turin, has a variant intended to maximize data flow to graphics processing units (GPUs) for quicker artificial intelligence computation.

AMD also introduced three new laptop PC processors based on the Zen 5 architecture that will work with Microsoft’s Copilot+ software and are optimized for AI applications.

In August, AMD agreed to pay $4.9 billion in cash and shares to purchase ZT Systems. Providing hyperscale server solutions for cloud computing, the latter is a general-purpose compute infrastructure and artificial intelligence supplier for hyperscale computing firms. The company is proud to have a worldwide manufacturing operation that spans the US, EMEA, and APAC.

The semiconductor industry is seeing high demand due to the expansion of AI applications, which coincides with AMD’s new objectives. As businesses expand their manufacture of AI-focused chips, supply chains are under stress due to the emergence of generative AI and sophisticated technologies. The surge in demand for AI chips has sparked worries about possible shortages.

The AI-driven spike in demand for GPUs alone may raise overall demand for some upstream components by 30% or more by 2026, according to research by Bain & Company. The industry may not be able to satisfy demand despite expenditures like the US CHIPS Act due to supply limitations and geopolitical conflicts, especially considering how difficult it is to scale production for sophisticated AI chips.