In collaboration withMicrosoft Azure and AMD
In Seattle, a weather scientist evaluates dynamic atmospheric models to foresee the upcoming major storm system. In Stuttgart, an automotive designer inspects crash-test simulations for vehicle safety accreditation. And in Singapore, a finance expert conducts simulated portfolio stress tests to safeguard against global economic disturbances.
Each of these experts—and the consumers, commuters, and investors who rely on their analyses—depends on a time-honored foundation of high-performance computing: the unassuming CPU.
With AI advancements powered by GPUs receiving the majority of press (and funding) in 2025, it’s easy to presume that CPUs are a thing of the past. Recent forecasts project that GPU and accelerator implementations will grow by 17% annually through 2030. However, in truth, CPUs remain crucial for the bulk of today’s most innovative scientific, engineering, and research tasks. Evan Burness, head of Microsoft Azure’s HPC and AI product teams, estimates that CPUs still enable 80% to 90% of HPC simulation tasks today.
In 2025, these systems are not only far from being outdated, but they are also undergoing a technological resurgence. A new surge of CPU advancements, including high-bandwidth memory (HBM), is yielding significant performance improvements—without the need for expensive architectural overhauls.
For further details, view the new webcast “Powering HPC with next-generation CPUs.“
This material was generated by Insights, the custom content division of MIT Technology Review. It was not crafted by the editorial team of MIT Technology Review. It was researched, designed, and written by human authors, editors, analysts, and illustrators. AI tools that may have been utilized were confined to secondary production processes that underwent thorough human scrutiny.
