AMD Acquires MEXT to Make Flash Perform Like DRAM

According to AMD, the company has officially acquired MEXT, a pioneer in AI-driven memory optimization technology, in a strategic move aimed at addressing one of the most pressing challenges facing modern data centers: the growing gap between memory demand, performance requirements, and infrastructure costs. As AI models, data analytics platforms, and high-performance computing (HPC) workloads continue to expand rapidly, memory capacity and efficiency have emerged as critical bottlenecks limiting scalability and profitability.

Founded in 2023 and headquartered in Santa Clara, California, MEXT developed a software-based technology known as the Predictive Memory Engine™. The platform uses artificial intelligence to analyze memory access patterns in real time, automatically relocating infrequently accessed “cold” data to lower-cost NAND flash storage while predicting and preloading frequently accessed “hot” data into DRAM before it is needed.

This approach enables flash storage to deliver performance characteristics much closer to DRAM, allowing servers to significantly expand available memory capacity without requiring proportional increases in expensive memory hardware. AMD stated that the technology can effectively increase usable memory capacity by two to four times while maintaining application performance levels. In large-scale deployments, this could reduce memory hardware costs by as much as 50%.

The acquisition comes at a time when emerging workloads such as Agentic AI, large language models (LLMs), and advanced analytics are driving unprecedented memory requirements. At the same time, fluctuations in DRAM pricing and supply continue to challenge cloud operators and enterprise infrastructure providers. By integrating MEXT’s software-defined memory technology into its data center portfolio, AMD aims to provide customers with a more cost-efficient path to scaling AI infrastructure.

Dan McNamara, Senior Vice President and General Manager of Computing and Enterprise AI at AMD, said the integration of MEXT’s technology is expected to help organizations maximize the value of their infrastructure investments while accelerating AI deployment across data center environments.

Industry analysts view the acquisition as an important step in AMD’s strategy to strengthen its end-to-end AI ecosystem. The addition of MEXT enhances AMD’s existing portfolio, which already includes EPYC server processors, Instinct accelerators, and the ROCm software platform. By adding an advanced memory optimization layer, AMD can offer a more comprehensive infrastructure stack designed to improve performance, efficiency, and total cost of ownership.

Jay Goldberg, an analyst at Seaport Research, noted that even modest improvements in memory efficiency can generate substantial savings in hyperscale computing environments worth tens of billions of dollars. He emphasized that the transaction is primarily intended to improve AMD’s competitiveness in data center infrastructure rather than signal any entry into memory manufacturing.

Investors responded positively to the announcement. AMD shares surged nearly 7% on June 15, closing at $547.26. The company’s market capitalization briefly exceeded $900 billion during trading before ending the session at approximately $892 billion, surpassing JPMorgan. The rally was also supported by broader gains across AI and semiconductor stocks, as well as a recent upgrade from Citigroup, which raised AMD’s rating to “Buy” and set a $575 price target.

Benchmark analyst Cody Acree stated that the acquisition is unlikely to have a significant impact on AMD’s near-term revenue. However, he described the deal as a valuable technology enhancement that broadens the company’s memory architecture capabilities and strengthens its long-term positioning in AI infrastructure.

As large language models advance toward trillion-parameter scales, simply adding more DRAM is becoming increasingly impractical due to cost, power consumption, and physical space limitations. AMD’s acquisition of MEXT reflects a broader industry shift away from relying solely on hardware upgrades and toward software-defined memory architectures enhanced by AI-driven optimization.

By leveraging MEXT’s technology, AMD gains a powerful tool for expanding effective memory capacity, lowering total cost of ownership (TCO), and improving resource utilization without substantial hardware investments. The acquisition is expected to strengthen AMD’s competitive position in AI infrastructure while encouraging the broader semiconductor industry to rethink traditional memory hierarchy and data center architecture strategies.