API
< Back to all clusters
[TECHNOLOGY] · United States · 4 sources

started · updated

AI infrastructure demands massive investment and memory innovations

The rapid growth of artificial intelligence is driving a massive demand for computing capacity and memory resources. According to the Bain and Company 2026 Global Technology Report, building approximately 150 GW of new capacity by 2030 will require investments between $5 trillion and $6.5 trillion. This scale represents a tripling of global capacity in just five years, with individual data center campuses now projected to reach 5 GW or more, costing up to $200 billion per installation.

To sustain this level of investment, the AI market would need to generate roughly $6 trillion in annual revenue by 2031. While current enterprise and consumer applications are expected to contribute $1.2 trillion to $1.8 trillion, an additional $4.2 trillion must be generated through new value sources and innovations.

Addressing the high cost and scarcity of RAM, AMD has developed MEXT, a software technology designed to extend memory capacity. MEXT uses artificial intelligence to predict data access patterns, allowing less frequently used memory pages to be moved to flash storage while preemptively retrieving necessary data. This approach aims to reduce the reliance on expensive DRAM and lower infrastructure costs without incurring the performance penalties typically associated with flash memory latency.

Entities

AMD · Bain and Company