The great flip: how accelerated computing redefined scientific systems and enabled Artificial Intelligence at scale

Over the past 15 years GPUs and accelerated computing have reshaped supercomputing, making Artificial Intelligence performance the new yardstick for scientific systems. The shift began as a power-efficiency imperative and evolved into coordinated design across processors, networking and software.

What the article calls the great flip describes how graphics processing units born from gaming have moved upstream to transform supercomputing and to carry Artificial Intelligence into the most advanced scientific systems. JUPITER at Forschungszentrum Jülich is presented as an emblem of this era, delivering 63.3 gigaflops per watt and 116 Artificial Intelligence exaflops, up from 92 at ISC High Performance 2025. The composition of leadership systems has changed rapidly: in 2019 nearly 70 percent of TOP100 systems were CPU-only, a figure that has fallen below 15 percent today; 88 of the TOP100 are now accelerated and about 80 percent of those use NVIDIA GPUs.

Across the broader TOP500, the article reports that 388 systems, or 78 percent, now use NVIDIA technology, including 218 GPU-accelerated systems, an increase of 34 systems year over year, and 362 systems connected by NVIDIA networking. The piece highlights architectures such as Hopper and Blackwell and the CUDA-X platform as key enablers that let scientific systems operate across precision modes from FP64 to mixed precision FP32 and FP16 and extremely efficient integer formats like INT8. That flexibility and efficiency is framed as the reason researchers can stretch power budgets to run larger simulations and train deeper models, with JUPITER offering both 116 Artificial Intelligence exaflops and 1 EF FP64 as a concrete example of simulation and Artificial Intelligence working together.

The article traces the technical arc through milestone systems-Titan in 2012, Piz Daint in 2013, and Summit and Sierra in 2017-that shifted the field to acceleration first. It cites energy and performance metrics such as the Green500 list, where the top eight systems are NVIDIA-accelerated and NVIDIA Quantum InfiniBand connects seven of the top ten. The narrative includes Jensen Huang’s early prediction at SC16 that deep learning would reshape leadership computing. The conclusion is that a power-efficiency imperative became an architectural advantage and has matured into a scientific superpower, enabling advances in climate modeling, drug discovery, fusion and quantum simulation as simulation and Artificial Intelligence scale together.

70

Impact Score

YouTube to automatically label Artificial Intelligence-generated videos

YouTube is shifting from voluntary disclosure to automated detection for significant photorealistic Artificial Intelligence-generated video content. Labels will become more visible across long-form videos and Shorts, with permanent markers for content made with YouTube tools or verified through provenance systems.

Axiom Math says its proofs reached peer reviewed journals

Axiom Math says proofs generated by its system have been accepted by several peer-reviewed journals, pairing machine-checkable formal proofs with human-authored papers. The development adds evidence that Artificial Intelligence tools are beginning to contribute to publishable mathematical research.

Google expands Gemini for Science

Google is rolling out Gemini for Science, a set of experimental tools aimed at compressing scientific work that would typically take months or years into days. The effort combines multi-agent research systems, computational discovery tools, literature analysis, and database-connected life science assistants.

Europe weighs technology sovereignty push amid internal debate

Europe is preparing a new policy push to reduce reliance on major technology platforms, but internal disagreements are shaping the scope and pace of the effort. The Artificial Intelligence Development Act is due to be unveiled on June 3 after repeated delays.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.