On November 16, 2026, at the Supercomputing 2026 conference in Chicago, Altera Corporation, Oxford Quantum Circuits, and NVIDIA unveiled an FPGA-based quantum system controller reference design that couples quantum processors directly to GPU computing. Built on Altera Agilex 9 Direct RF FPGAs and NVIDIA NVQLink, the architecture replaces discrete analog racks with unified, low-latency control and machine-learning readout.
Joint Unveiling at Supercomputing 2026 in Chicago
The announcement, made at Supercomputing 2026 (SC26) in Chicago on November 16, 2026, addresses a central physical bottleneck in experimental quantum systems: the volume and footprint of classical control electronics. As qubit counts grow, traditional control equipment expands into multi-rack assemblies filled with separate signal generators, mixers, filters, and standalone converters.
To solve this density challenge, Altera Corporation (the standalone FPGA provider majority-owned by Silver Lake, with Intel holding 49%), Oxford Quantum Circuits (OQC), and NVIDIA created an integrated controller reference design. Demonstrated live at Altera’s SC26 booth (#2854), the system runs real-time qubit control, spectroscopy, and low-latency readout between an Altera Agilex 9 Direct RF FPGA and a next-generation OQC superconducting quantum processing unit (QPU), with high-throughput connectivity to NVIDIA GPUs via NVIDIA NVQLink.
Technical Architecture: Agilex 9 Direct RF and NVQLink Integration
Superconducting qubits require precise microwave pulses in the 4 GHz to 8 GHz band. In standard setups, digital logic generates intermediate-frequency signals routed through external analog mixers, oscillators, and filters before reaching the dilution refrigerator, with mirrored chains on the readout path.
The Altera Agilex 9 Direct RF FPGA replaces those external conversion stages by embedding high-speed ADCs and DACs operating at direct RF sampling rates up to 64 gigasamples per second (GSPS). Direct synthesis and digitization in the digital domain eliminate analog up- and down-conversion, reducing electrical drift, thermal load, and component counts. Frequency-domain multiplexing enables multiple qubit control and resonator readout channels to share single physical RF ports, allowing researchers to execute a complex quantum circuit without sprawling coaxial cabling.
NVIDIA NVQLink connects the Agilex 9 FPGA directly to NVIDIA accelerated computing with sub-4-microsecond data-transfer latencies across high-bandwidth switch channels. This architecture establishes a strict division of labor: the deterministic FPGA manages time-critical pulse generation, synchronized gate scheduling, and resonator spectroscopy, while GPUs execute compute-heavy calibration, state tomography, and machine-learning readout classification. Through CUDA-Q libraries, developers can run GPU kernels that dynamically steer FPGA registers during execution of a hybrid quantum algorithm.
System Comparison: Traditional Discrete Racks vs. Integrated Reference Controller
The transition from discrete RF racks to an integrated Direct RF FPGA and GPU-linked controller impacts footprint, latency, and operational flexibility:
| System Feature | Traditional Discrete Instrument Racks | Altera-OQC-NVIDIA Reference Design |
|---|---|---|
| RF Signal Synthesis | Baseband/IF DACs requiring external analog mixers and oscillators | Direct RF sampling up to 64 GSPS on Agilex 9 Direct RF FPGA |
| Hardware Density | Multiple multi-chassis 19-inch racks for 20 to 50 physical qubits | Single-chassis FPGA board with multi-channel frequency multiplexing |
| QPU-to-GPU Interconnect | PCIe or Ethernet host links with millisecond-scale latency | NVIDIA NVQLink providing sub-4-microsecond data transfer |
| Readout Classification | Static thresholding on FPGA or delayed offline post-processing | Real-time machine-learning classification on NVIDIA GPUs |
| Software Stack | Vendor-specific proprietary scripts and fragmented APIs | Unified CUDA-Q stack with built-in hardware emulation modes |
Background: The Quantum-Classical Scaling Dilemma
As quantum hardware developers advance toward logical qubits and fault tolerance, classical control hardware has become a primary physical barrier to scaling. Housing dozens of chassis and routing hundreds of coaxial lines into dilution refrigerators creates severe heat loads, cable phase drift, and high energy costs, mirroring challenges seen across installations like the IBM Quantum Platform.
Altera’s position reflects its September 12, 2025 separation from Intel, in which Silver Lake acquired a 51% majority stake while Intel retained 49%. As an independent pure-play FPGA firm, Altera has targeted embedded RF processing and AI acceleration. Concurrently, Oxford Quantum Circuits has focused on enterprise data center deployments, operating its GENESIS quantum hardware in facilities such as Digital Realty’s JFK10 site in New York, where hybrid quantum-GPU infrastructure is essential for enterprise modeling.
Stakeholder and Executive Perspectives
Executives across Altera, OQC, and NVIDIA stressed that standardizing controller reference designs is necessary to transition quantum testbeds into production computing infrastructure.
Farhad Shafai, Head of Business Solutions and Marketing at Altera, described the practical significance of the joint design: “Our work with Oxford Quantum Circuits and NVIDIA uses FPGAs to demonstrate a full system reference design in a real-world setting, from the quantum processor through the FPGA to the GPU. This demonstration shows how Altera’s Direct RF technology can anchor the control layer of quantum systems, and it gives developers a clear path to build on.”
Simon Phillips, Chief Product Officer at Oxford Quantum Circuits, emphasized deterministic feedback for scaling qubit performance: “Precise control and fast feedback are essential to unlocking the capabilities of our quantum processors. Altera’s FPGA technology provides a programmable, low-latency connection to our QPU, bringing control and readout together while allowing us to refine calibration and feedback as our systems evolve. This integration helps simplify system development and provides a path to supporting more qubits.”
Sam Stanwyck, Director of Quantum Product at NVIDIA, noted the impact of open GPU connectivity: “Advances in AI are opening new ways to build and operate quantum computers. NVIDIA NVQLink gives quantum builders an open architecture to put those advances to work. Together with Altera and OQC, we’re giving developers a working reference design to explore AI-driven calibration and readout and accelerate the development of their own quantum systems.”
Implications for Researchers, Universities, and Research Offices
The introduction of unified FPGA-and-GPU reference controllers carries practical implications for academic institutions, national laboratories, and research administration:
- Laboratory Footprint and Infrastructure Budgets: University labs face tight physical space and power budgets. Replacing racks of analog signal generators with integrated Direct RF FPGAs reduces cleanroom footprint, cooling requirements, and electrical utility demands.
- Pre-Hardware Development via Emulation: For research groups lacking dedicated dilution refrigerators or continuous QPU access, the package includes an FPGA emulation mode. Researchers and students can calibrate pulse sequences and test neural network readout classifiers before reserving scarce physical hardware.
- Export Control and Compliance Oversight: University research offices must track regulatory controls on advanced Direct RF FPGAs. Devices supporting direct RF synthesis up to 64 GSPS intersect with dual-use export regulations and deemed-export rules, as outlined in guidance for quantum computing export controls and foreign-national access. Compliance teams must maintain appropriate lab security and citizenship access protocols.
- Unified Software Stacks for Interdisciplinary Teams: By coupling CUDA-Q with standard GPU workflows, the reference design lowers entry barriers for computer scientists and mathematicians, enabling quantum-classical orchestration via Python and C++ without requiring custom VHDL logic design.
Commercial Rollout and Next Steps
The complete controller reference design is on display throughout SC26 at Altera’s booth in Chicago. Altera, OQC, and NVIDIA have scheduled the formal commercial release of the reference hardware design and software package for January 2027.
Following the January 2027 release, OQC plans to integrate the Direct RF control architecture into upcoming commercial QPU deployments, providing enterprise clients with native NVQLink connectivity to GPU clusters. Concurrently, NVIDIA continues deploying NVQLink across global supercomputing centers, establishing the reference architecture as a model for hybrid high-performance computing centers.
Sources
- The Quantum Insider: Altera, OQC and NVIDIA Link Quantum Processor to GPU Computing, November 16, 2026
- StorageReview: Altera Agilex 9 Direct RF FPGA Anchors a QPU-to-GPU Quantum Controller Design With OQC and NVIDIA NVQLink, November 16, 2026
- MLQ.ai: Altera, OQC and Nvidia build an FPGA controller design linking a quantum processor to GPUs, November 16, 2026
- Oxford Quantum Circuits: OQC Advances Integration of Quantum Computing and AI Supercomputing with NVIDIA NVQLink, November 16, 2026
- NVIDIA Newsroom: World’s Leading Scientific Supercomputing Centers Adopt NVIDIA NVQLink to Connect QPUs and GPUs, November 16, 2026







