AI & ComputeLOT 261122 · 3 min read
Chinese Team Wins 2026 Gordon Bell Prize, Cracks FeMoco at 1.456 Exaflops
A Chinese team won the 2026 ACM Gordon Bell Prize for simulating quantum chemistry's flagship FeMoco molecule at 1.456 double-precision exaflops on the LineShine system, challenging quantum-advantage claims.
Key takeaways
- A 10-member Chinese team won the 2026 ACM Gordon Bell Prize at SC26 in Chicago on Nov. 19, 2026.
- The code sustained 1.456 exaflops in double precision (58.9% of peak) and 5.626 exaflops mixed-precision on 20,480 LineShine nodes.
- The work simulates FeMoco, the benchmark problem for quantum computers since a 2016 Microsoft–ETH Zurich paper in PNAS.
- The team is led by Honghui Shang of USTC in Hefei; the prize carries a $10,000 award funded by Gordon Bell.
A 10-member Chinese team sustained 1.456 exaflops in double precision — 58.9% of peak — while simulating FeMoco, the molecule researchers have spent a decade using as the flagship problem for quantum computers. ACM awarded the team the 2026 Gordon Bell Prize on Thursday, Nov. 19, at the SC26 conference in Chicago, supercomputing's most recognizable honor.
The winning project is titled "Closing the FeMo-Cofactor 'Quantum-Advantage' Gap with Exascale Supercomputers." ACM credited the team with the first exascale classical solution to the FeMoco CAS problem — a model of the electronic structure of the iron-molybdenum cofactor inside nitrogenase, the enzyme that converts atmospheric nitrogen into ammonia.
What hardware ran it?
According to the paper's abstract, quoted by ACM, the code was tuned for two Chinese exascale systems: LineShine and China's New-generation Intelligent Supercomputer, or CNIS. On 20,480 LineShine nodes, the code held 1.456 EFLOP/s in double precision and reached 5.626 EFLOP/s in mixed precision. An exaflop is a billion billion floating-point operations per second.
ACM gave no further specifications for either machine. That silence is itself notable; China's exascale systems remain largely undocumented in public TOP500 submissions, and the prize citation is one of the few official confirmations of their sustained performance.
Honghui Shang of the University of Science and Technology of China (USTC) in Hefei leads the team. Its members come from USTC, the Chinese Academy of Sciences, the National High Performance Computer Engineering Technology Research Center in Beijing, and Sun Yat-sen University in Shenzhen.
Why FeMoco?
FeMoco earned its celebrity status in 2016, when researchers at Microsoft and ETH Zurich published a paper — later appearing in PNAS — describing how an error-corrected quantum computer could resolve the nitrogen fixation mechanism. They argued classical methods couldn't reach the required accuracy at reasonable cost. The problem has since anchored many published resource estimates for future quantum hardware.
The 2026 winning project's title frames the work squarely against that claim: classical exascale machines narrowing the gap quantum proponents had declared unbridgeable. A $10,000 award, funded by Gordon Bell, accompanies the prize.
There are limits to what the result proves, though. ACM's release doesn't say how close the simulation comes to the accuracy quantum chemists want for the full nitrogen fixation reaction mechanism, nor how it stacks up against published quantum resource estimates. So the prize doesn't, by itself, settle whether FeMoco remains a valid quantum-advantage target.
A crowded week for quantum claims
The award landed during a week dense with competing quantum milestones. Google published contextuality results on its Willow chip in the same window, part of the running argument over when — or whether — quantum hardware outperforms classical machines on real chemistry.
The Gordon Bell result sharpens that debate from the classical side. Every time a classical supercomputer absorbs a problem once reserved for quantum machines, quantum vendors must push their claimed crossover point further out — and their resource estimates must account for faster classical baselines.
Whether FeMoco stays on the quantum roadmap as a genuine advantage target will depend on the accuracy details ACM's citation leaves out, and on whether classical exascale codes like this one keep closing the gap as Chinese systems scale.
mlq.ai (Source)
More from Rebecca Stone
Show full bio
Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,245 articles