DARPA Launches HARQ to Build Quantum Computers That Mix Qubit Technologies
DARPA's Heterogeneous Architectures for Quantum program will fund 19 teams over 24 months to combine different qubit types into unified quantum systems, abandoning the single-technology approach.
Overview
The Defense Advanced Research Projects Agency launched its Heterogeneous Architectures for Quantum (HARQ) program on April 14, 2026, funding 19 performer teams across 15 organizations to develop quantum computers that combine multiple qubit technologies into a single unified system. The initiative marks a departure from the prevailing industry approach of building entire quantum platforms around a single qubit type, a constraint that DARPA argues has become a structural ceiling on the field’s progress.
What We Know
HARQ is organized into two parallel workstreams operating over a 24-month development period. The first, called MOSAIC (Multi-qubit Optimized Software Architecture through Interconnected Compilation), focuses on the software layer: building circuit compilers and optimization frameworks that can route quantum operations to whichever qubit type is best suited for the task. Performers on the MOSAIC track include Infleqtion, memQ, Q-CTRL, the University of Michigan, and the University of Pennsylvania, according to DARPA’s program announcement.
The second workstream, the Quantum Shared Backbone (QSB), addresses the harder hardware problem: enabling high-fidelity communication between physically distinct qubit platforms within a single system. QSB performers include IonQ, Harvard University, Stanford University, UC Berkeley, the University of Illinois Urbana-Champaign, Carnegie Mellon University, the Australian National University, EPFL, and the University of Texas at Austin, as reported by The Quantum Insider.
DARPA Program Manager Justin Cohen, of the Microsystems Technology Office, framed the initiative as a deliberate challenge to the field’s assumptions: “Qubit technologies each have their own distinct advantages, but no single approach can deliver everything needed for large-scale, high-performance quantum systems,” he stated, according to DARPA. Cohen described the goal as moving past “a one-qubit-to-rule-them-all mindset.”
IonQ’s role in the QSB workstream centers on quantum memories fabricated from quantum-grade synthetic diamond. The company’s interconnect technology is designed to link trapped-ion qubits with other architectures — including neutral atoms and superconducting qubits — over both short datacenter-scale distances and longer fiber-optic spans. IonQ CEO Niccolo de Masi said the company’s “pioneering quantum interconnect technology can enable modular scalability not only for ion traps, but for a wide range of quantum technologies,” as stated in IonQ’s announcement. IonQ noted it achieved 99.99% two-qubit gate fidelity and completed the first qubit-to-photon frequency conversion compatible with standard fiber optic infrastructure in 2025, achievements that underpin its HARQ role.
memQ Chief Product Officer Manish Singh stated that “the HARQ program will catalyze the modularity, scale, and resource optimization needed” for quantum computing to reach practical utility, according to the Chicago Quantum Exchange.
The DARPA program page describes the potential upside in stark terms: compiler tools that route operations across qubit types could reduce resource requirements by a factor of 1,000 compared to single-technology systems. The agency also plans a government-led architecture study examining software-hardware co-design, scalability, and economic implications alongside the performer work.
The Case for Mixing Qubit Types
The premise of HARQ rests on a long-standing observation in quantum hardware: different qubit technologies excel at different things. Superconducting qubits, used by IBM and Google, operate at nanosecond speeds but require dilution refrigerators and are difficult to network between separate chips. Trapped-ion qubits, used by IonQ and Quantinuum, offer exceptional fidelity and long coherence times but operate more slowly. Neutral-atom qubits, used by QuEra and Pasqal, can be reconfigured mid-computation and have recently entered error-correction regimes. Photonic qubits are well-suited for communication but poorly suited for local computation.
No single platform dominates across all of these axes. Classical computing resolved an analogous tension by building heterogeneous systems — CPUs handle sequential logic, GPUs handle parallelism, and ASICs handle specialized workloads — all sharing memory and connected by high-speed buses. HARQ is an explicit attempt to determine whether the same architectural strategy can be applied to quantum systems.
The core engineering challenges are substantial. Qubits of different types operate at different frequencies, temperatures, and timescales. Connecting them without destroying coherence requires quantum transducers and interconnects that do not yet exist in production-ready forms. HARQ’s QSB workstream is intended to produce those components.
What We Don’t Know
DARPA has not disclosed total funding for the program. The 24-month timeframe suggests HARQ is a research and feasibility effort rather than a near-term engineering program; whether it produces field-deployable hardware or primarily generates architectural principles and software tools remains to be seen.
The performer list covers a wide range of qubit modalities — trapped ions, neutral atoms, superconducting, and photonic — but it is not yet clear how many distinct combinations the teams will attempt to interconnect, or what performance thresholds DARPA has defined for success.
Federal quantum investment has accelerated since the 2018 National Quantum Initiative, and HARQ fits within a broader U.S. strategy to maintain leadership in quantum hardware. How HARQ intersects with parallel industry efforts — including IBM’s fault-tolerance roadmap, Google’s surface-code error correction work, and Microsoft’s topological qubit program — will determine whether the heterogeneous approach gains industry traction or remains a government research track.
Analysis
The HARQ announcement is notable less for the hardware it will immediately produce than for the architectural assumption it encodes. By funding 19 teams simultaneously under a shared framework, DARPA is effectively placing a research bet that the path to practically useful quantum computers runs through specialization and integration rather than through perfecting any single qubit type.
This mirrors a debate that ran through classical computing in the 1990s, when RISC architectures competed with CISC and x86 eventually prevailed not through architectural purity but through iterative fabrication improvements. Quantum computing has not yet reached its equivalent of the fabrication scaling era, but HARQ suggests that the U.S. defense research apparatus does not believe it needs to wait for one dominant qubit technology to emerge before beginning serious systems-level engineering.