AMD HPC executive Scott Tease says AI won't replace traditional supercomputing
Ahead of SC26, AMD's corporate vice president of HPC and sovereign AI, Scott Tease, told Tom's Hardware that most high-performance computing sites believe AI will not supplant HPC, and that a heterogeneous mix of HPC, AI and quantum computing is the better path forward. He pointed to Nvidia's Rubin GPU delivering 33 TFLOPs of FP64 compute, lower than the four-year-old H100's 34 TFLOPs, while AMD's Instinct MI355X reaches 78.6 TFLOPs and its upcoming MI430X targets 288 TFLOPs of FP64.
GoKawiil's interpretation of the reporting above, not reported fact.
The comments highlight a widening divergence between chipmakers optimizing for low-precision AI throughput and the high-precision FP64 math still required for scientific simulation and research. AMD's framing suggests it sees an opening to position its hardware as better suited to HPC workloads even as Nvidia emphasizes AI-oriented performance gains. The remarks also reflect AMD's involvement in heterogeneous computing efforts, such as its pilot with IBM at Oak Ridge National Laboratory, which could shape how future supercomputing systems are built.
- AMD's Scott Tease argues AI will not replace traditional HPC workloads that require high precision.
- Nvidia's Rubin GPU has lower native FP64 performance than its four-year-old H100, while AMD's chips show higher FP64 throughput.
- AMD is piloting a heterogeneous compute stack combining HPC, AI and quantum computing with IBM at Oak Ridge National Laboratory.
Source: tomshardware.com — Jake Roach, 2026-11-16
Published there as: “Researchers ‘recognize that AI will never replace HPC,’ says AMD executive — high-precision workloads will ‘exist for a very long time’ even as AI usage expands”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.