IonQ (IONQ), one of the most closely watched publicly traded quantum-computing companies, took a hit in January 2025 after Nvidia (NVDA) CEO Jensen Huang questioned how quickly useful quantum machines would arrive.
Huang said a reasonable estimate for very useful quantum computers was 15 to 30 years. IonQ shares fell 9.4% that day.
Less than two years later, Nvidia is preparing to put an IonQ quantum processor inside its own research center.
IonQ said Sept. 23 that Superion 256 will become the first on-premise quantum processor at Nvidia’s Accelerated Quantum Research Center, or NVAQC. The system is scheduled to be installed in 2027 and connected directly to an Nvidia GB200 NVL72 system.
IonQ shares were up about 4.98% to $42.77 around midday Sept. 23, after trading as high as $46.02 earlier in the session.
A day earlier, IonQ also announced a real-time quantum error-correction decoder that runs on a conventional CPU.
The two announcements show how IonQ is building around a system in which quantum processors work with CPUs and GPUs inside the same computing environment.
Nvidia will connect IonQ’s QPU directly to Blackwell
IonQ’s Superion 256 is scheduled to be installed at NVAQC in 2027 and directly linked to an Nvidia GB200 NVL72 through Nvidia’s NVQLink technology.
The workloads will run through CUDA-Q, Nvidia’s software platform for coordinating quantum and conventional computing. A quantum processing unit, or QPU, performs calculations using quantum bits, or qubits.
Related: Binance bets $100 million on Circle. Watch this margin
In the planned Nvidia setup, the processors have different jobs.
IonQ’s QPU will perform quantum calculations. Nvidia’s GPUs will handle accelerated conventional computing, with CUDA-Q coordinating work across the systems.
Nvidia designed NVAQC around this type of architecture.
When the company announced the Boston research center in March 2025, Nvidia said the facility would integrate quantum hardware with AI supercomputers to work on problems such as qubit noise, error correction, and hybrid algorithms.
Superion 256 will be the first QPU installed on-site.
Nvidia moved from quantum skepticism to building infrastructure
Nvidia’s research-center deployment arrives less than two years after Huang’s long commercialization timeline helped trigger a selloff across quantum stocks.
By March 2025, Nvidia had already moved from discussing the timeline to building hardware and software around the technology.
NVAQC was announced with partners including Quantinuum, Quantum Machines, and QuEra, as well as researchers from Harvard and MIT. Nvidia equipped the center with AI supercomputing infrastructure designed to work with multiple types of quantum hardware.
More AI:
- Nvidia just made a move Wall Street wasn’t ready for
- Microsoft just took sides in AI policy fight
- OpenAI just disclosed something genuinely alarming
But no evidence has shown that Nvidia expects large-scale quantum computers to replace conventional supercomputers soon. IonQ is also not Nvidia’s only quantum collaborator.
The new step is narrower and more concrete: IonQ will be the first company to install an on-premise QPU at NVAQC.
Nvidia is now building the infrastructure required to test quantum processors beside its GPUs rather than simply debating when useful quantum computing will arrive.
IonQ’s CPU decoder tackles the error-correction bottleneck
Qubits are highly sensitive to noise, which can introduce errors during a quantum calculation.
A fault-tolerant quantum computer must detect and correct those errors continuously.
The correction process relies on conventional computing hardware. If the classical decoder cannot process error information fast enough, the quantum processor may have to pause.
IonQ said Sept. 22 that it developed and tested an error-correction decoder that can run continuously on one standard off-the-shelf CPU.
What IonQ tested
- Up to 408 logical qubits in simulated benchmark circuits.
- More than 31.5 million quantum operations across those circuits.
- As little as 0.02% added execution time, which IonQ calls “stretch” time.
Those figures need one important qualification. IonQ did not demonstrate a physical quantum computer running 408 logical qubits. The company tested its decoder on benchmark circuits simulating that scale.
The experiment tested whether a conventional processor could keep pace with the error-correction workload.
That fits the architecture IonQ plans to test with Nvidia: conventional processors handle the supporting computation around a QPU.
Nvidia gives IonQ a showcase: 2027 deliveries are the commercial test
IonQ began accepting orders for Superion 256 on Sept. 8.
The company expects first customer deliveries in 2027 and said it had pre-sold its first Superion 256 system during the first quarter.
IonQ has not disclosed the financial value of the NVAQC deployment.
Its latest financial figures also show a company growing rapidly and spending heavily as it expands both quantum technology and manufacturing.
IonQ’s business by the numbers
- $80.1 million: Second-quarter revenue, up 287% year over year.
- $120.3 million: Adjusted EBITDA loss for the quarter.
- About $2.0 billion: Cash, cash equivalents and investments on a pro forma basis after the SkyWater acquisition.
- $450 million to $460 million: IonQ’s latest 2026 revenue guidance.
IonQ’s earlier $280 million to $290 million outlook excluded SkyWater. The company raised its forecast to $450 million to $460 million after closing the acquisition, and the new range includes SkyWater revenue from July 31 through year-end.
So the higher forecast does not represent organic quantum-computing growth alone.
Superion provides a separate test of IonQ’s core quantum hardware business.
The Nvidia installation is scheduled for next year. Customer deliveries are also expected to begin in 2027, followed by IonQ’s target for commercially manufacturable fault-tolerant systems in 2028.
The next concrete proof will come when Superion starts shipping to paying customers in 2027, and those deployments begin generating repeat orders and revenue.
Related: Microsoft is killing an Office app you’ve used for decades
