AMD has acquired MEXT, adding AI-driven memory optimization technology to its data center and AI portfolio. The deal targets a practical bottleneck in modern compute infrastructure: memory capacity and efficiency constraints as AI, data analytics, virtualization, and high-performance computing workloads continue to grow.

MEXT’s core technology is described as AI-powered predictive memory designed to make flash behave more like DRAM. The stated goal is to expand usable memory capacity while maintaining performance and efficiency, which AMD links to reducing infrastructure costs, improving resource utilization, and accelerating deployments for general-purpose and AI workloads.

For data center operators, memory pressure shows up in several places: larger model footprints, bigger working sets, and more contention as consolidation increases. Any approach that can increase effective memory capacity without a proportional increase in DRAM spend can change server sizing and node counts. But the engineering reality is that “flash like DRAM” hinges on how well prediction and caching can mask flash latency under real workload mixes.

AMD said it plans to integrate MEXT’s technology across the AMD data center portfolio, and it expects the combination to help enterprise customers “unlock greater value from their infrastructure investments while accelerating AI deployment.” AMD also highlighted that the acquisition brings in a team with expertise in memory systems and AI infrastructure.

Source: AMD