IBM and Nvidia tap Fermilab to stitch modular quantum computers together
IBM, Nvidia and Fermilab unveiled plans for a co-design platform pairing IBM's 120-qubit Nighthawk chip with Nvidia's DGX B200 machines. The goal: crack error correction and figure out how to link quantum processors into bigger systems. Nvidia also dropped a new decoding algorithm it says can slash some computation times by up to 13,000x.
- 6 sources report it: cryptobriefing.com, breakingthenews.net, u.today, crypto-economy.com and 2 more.
- The source published it on Nov 16, 2026 at 4:21 PM.
Nothing open. Every check passed.
The three partners announced the plan on November 16, 2026. IBM's Quantum Nighthawk processor handles the quantum side, and Nvidia's DGX B200 systems handle the classical side. Fermilab, through its Superconducting Quantum Materials and Systems Center, does the testing. The stated targets are error correction, scalable architectures and scientific applications, with high-energy physics and nuclear physics as the first fields in line.
Nighthawk carries 120 qubits. IBM says it runs circuits holding up to 7,500 two-qubit operations, hitting speeds of up to 100,000 samples per second. The DGX B200 machines on the other end are the same class of accelerated computers that chew through large AI workloads, now pointed at quantum chips instead.
Fermilab's SQMS Center gets a specific job: test and develop ways to wire multiple quantum processors together, and scale up the classical resources as the quantum side grows. That interconnect problem is the whole reason modular quantum computing is hard. The center also gets real scientific users attached to the hardware.
Nvidia used the announcement to roll out a decoding algorithm called TCBP-JOBS, aimed at quantum error-correction research. Nvidia says it can shrink certain computation times from 12 weeks down to 45 minutes, a speedup of up to 13,000x. Both qualifiers matter here: 13,000x is the upper bound, and the 12-weeks-to-45-minutes number applies to certain tasks only.
The collaboration leans on ties that already exist. IBM's partnership with Fermilab's SQMS Center was formalized in 2024, and Nvidia has been building out CUDA-Q, with the CUDA-Q Logical release arriving in September 2026. IBM was also linked earlier in November 2026 to a collaboration with Oak Ridge National Laboratory and AMD. For anyone tracking IBM or Nvidia: this is a plan for a platform, not a commercial launch, and the eye-catching numbers come with caveats attached.
TopicsIBM · Nvidia · Fermilab · Quantum Nighthawk · DGX B200 · SQMS Center · TCBP-JOBS · CUDA-Q
Written by AI from verified reporting.

