Quantum Computing Economics in 2026: Roadmaps, Modalities, and the Post-Quantum Migration

Where the four modalities stand in 2026 #

The quantum hardware field has compressed into four modalities with viable roadmaps: superconducting transmons (IBM, Google, Rigetti), trapped ions (IonQ, Quantinuum), neutral atoms (Atom Computing, QuEra, Pasqal), and photonics (PsiQuantum, Xanadu). A fifth approach, Microsoft's topological qubits based on Majorana zero modes, returned to scientific contention with a February 2025 Nature paper and the peer review controversy that followed. Each modality optimizes a different axis of the engineering tradeoff: gate speed, fidelity, connectivity, qubit count, and operating temperature.

Superconducting transmons remain the volume leader. IBM Heron R2 runs 156 fixed-frequency transmon qubits with two-qubit gate errors near 0.3 percent and tunable couplers on a heavy-hex topology. IBM Condor packed 1121 qubits onto a single chip in late 2023 as a yield proof of concept. Google's Willow, announced in December 2024, used 105 qubits to show the surface code crossing the fault-tolerance threshold: as code distance rose from 3 to 5 to 7, the logical error rate fell by roughly a factor of two per step, the first hardware demonstration that adding physical qubits reduces rather than amplifies error. Trapped ions anchor the high-fidelity end: Quantinuum H2 operates 32 ytterbium ions at a 2 to 1 logical encoding, and IonQ Tempo targets 64 algorithmic qubits. Atom Computing's 1180-qubit cesium array and reconfigurable Rydberg systems from QuEra and Pasqal have moved the neutral-atom modality faster than expected.

IBM's modular bet and the System Two architecture #

IBM has committed more publicly to a quantitative roadmap than any other vendor. The 2026 deliverable is Kookaburra at 1386 qubits, built as three coupled chips of 462 qubits each using L-couplers that link processor modules within a single dilution refrigerator. Beyond Kookaburra sits Blue Jay at 4158 qubits across multiple cryostats connected by M-couplers, microwave-frequency links crossing the cold-warm boundary. Quantum System Two, unveiled at the 2023 Quantum Summit, is the physical envelope: three dilution refrigerators sharing classical control infrastructure, with system count rising as IBM provisions Kookaburra and beyond.

IBM is betting that chip-level scaling will plateau before fault tolerance and that progress beyond a few thousand physical qubits requires interconnect rather than monolithic integration. Heron R2 is the workhorse for paying customers via IBM Quantum Network. Condor is a yield demonstration, not a product. The competing philosophy comes from PsiQuantum, whose Brisbane facility, jointly funded by the Australian federal and Queensland state governments at roughly 940 million Australian dollars, is sized to fabricate photonic chips at wafer scale. PsiQuantum targets one million photonic logical qubits by 2027, dependent on resource state generation, fusion-based quantum computing primitives, and its GlobalFoundries partnership. IBM scales toward fault tolerance through interconnect, while PsiQuantum bets that photonic loss budgets and wafer-scale manufacturing will leapfrog the superconducting roadmap.

Surface code, magic state distillation, and the threshold #

The unifying technical bar is the fault-tolerance threshold. The surface code requires physical gate error rates below roughly 1 percent (10 to the minus 2) to operate and benefits from rates near 0.1 percent (10 to the minus 3) to keep overhead tractable. Production hardware in 2026 sits between these bounds: IBM Heron R2 reports two-qubit errors near 0.3 percent, Google Willow demonstrated 0.14 percent worst-case two-qubit Pauli error during its threshold run, and Quantinuum H2 reports trapped-ion fidelities above 99.9 percent on gates that take milliseconds rather than nanoseconds.

Below threshold is necessary but not sufficient. Fault-tolerant computation also requires non-Clifford gates, typically the T gate, which surface codes cannot implement transversally. The standard solution is magic state distillation, which consumes large numbers of physical qubits to produce a high-fidelity ancilla. Recent work has cut distillation overhead by factors of 5 to 20, but T gate production remains the dominant cost. Breaking 2048-bit RSA still requires roughly 20 million noisy physical qubits at 0.1 percent error under standard surface code assumptions, a bar no vendor will reach this decade. What changed in 2024 and 2025 is that the surface code moved from theoretical to empirical: Willow demonstrated below-threshold scaling, and Quantinuum, IBM, and AWS have published similar milestones.

Public-market repricing and the commercial advantage gap #

Quantum equities went through a violent two-year cycle. IonQ traded above 35 dollars in late 2024, fell below 7 dollars during the 2025 correction, and has stabilized in a 7 to 15 dollar range against trailing revenue under 60 million dollars. Rigetti Computing, hit by cash burn warnings, trades well below its de-SPAC reference. Quantum Computing Inc. has cycled through repeated spike-decline patterns. D-Wave Quantum, which sells annealers rather than gate-model machines, has held up better on optimization customers but remains a small-cap.

The repricing reflects a hard fact: there is no commercial quantum advantage in production today. McKinsey's 2024 Quantum Technology Monitor and Boston Consulting Group's 2025 update agree that aggregate quantum hardware revenue is roughly 1.5 to 2 billion dollars per year globally, dominated by government and research customers. The benchmarks where vendors claim advantage, random circuit sampling and certain bosonic systems, do not correspond to commercial workloads. Optimization, materials simulation, and machine learning subroutines have been repeatedly matched by classical heuristics on GPU clusters. Strategos models a base case where commercial advantage on materials and chemistry arrives between 2028 and 2031, with revenue at 5 to 10 billion dollars annually by 2030. The DARPA Quantum Benchmarking Initiative will produce the most credible third-party data on advantage timelines.

National programs and the geopolitics of quantum #

Quantum has become a strategic technology category in the same sense as semiconductors. The United States National Quantum Initiative, signed in 2018 and funded at roughly 1.2 billion dollars across NIST, NSF, DOE, and DARPA, expired at the end of fiscal 2023 and awaits reauthorization, with the proposed Act setting roughly 2.7 billion dollars over five years. The CHIPS and Science Act layered on funding for DOE Quantum Information Science research centers. China's program, centered on the National Laboratory for Quantum Information Sciences in Hefei, draws cumulative public reporting estimates of 10 to 15 billion dollars across central, provincial, and academy lines, although the program spans civilian and defense budgets.

The European Union's Quantum Flagship is in its second decade, with roughly 1 billion euros committed and a 2024 pivot toward a Quantum Act framework tying procurement to industrial policy. The United Kingdom's National Quantum Strategy, refreshed in 2023, allocates 2.5 billion pounds over ten years. Australia's commitment to PsiQuantum's Brisbane plant is the largest single bet by a national government on a private quantum company.

The geopolitical contest is sharper on export controls than funding. The United States, the United Kingdom, the Netherlands, Japan, and France aligned through 2024 and 2025 on multilateral controls covering dilution refrigerators below specified base temperatures, photonic and microwave components, cryogenic test equipment, and certain quantum software. China retaliated in late 2024 by restricting outbound flows of quantum-relevant materials. Supply chain diligence for hardware vendors now resembles semiconductor export control compliance.

Post-quantum cryptography: from research to procurement #

The most consequential near-term economic effect of quantum computing is not fault-tolerant machines but the migration of public-key cryptography. NIST's August 2024 publication of FIPS 203 (ML-KEM, the Kyber-derived key encapsulation mechanism), FIPS 204 (ML-DSA, the Dilithium-derived digital signature), and FIPS 205 (SLH-DSA, the SPHINCS+ stateless hash-based signature) ended an eight-year selection and turned post-quantum cryptography into a procurement question. NIST also continued evaluation of FIPS 206 (FALCON-derived FN-DSA) and a fourth-round track including HQC for code-based diversity.

The migration timeline is longer and more expensive than the mathematics suggests. NSA's CNSA 2.0 timeline requires national security systems to begin transition by 2025 and complete migration by 2033, with software signing, firmware, web browsers, and networking equipment leading the queue. CISA, NIST, and NSA jointly published a 2023 quantum readiness roadmap pointing federal agencies toward inventory, prioritization, and phased replacement. Strategos estimates federal PQC migration alone will run into the high single-digit billions of dollars, with regulated financial services and healthcare adding multiples. The harvest-now-decrypt-later threat model is the binding rationale: adversaries are presumed to be capturing encrypted traffic today against future quantum machines, and data with decades-long confidentiality requirements is exposed if migration lags.

Two scenarios for 2026 to 2030 #

Promethean's base case, at roughly 60 percent probability, has the field reaching fault tolerance on a small number of logical qubits across multiple modalities by 2027 and 2028, scaling to hundreds of logical qubits by 2030. IBM ships Kookaburra and a meaningful fraction of Blue Jay, Google delivers on its million-physical-qubit milestone, Quantinuum and IonQ keep closing the algorithmic-qubit gap, and Atom Computing or PsiQuantum produce a credible logical-qubit milestone outside the superconducting cluster. Aggregate quantum revenue grows to 5 to 10 billion dollars per year by 2030. PQC migration runs to schedule for federal systems and slips two to four years for the broader regulated economy.

The downside case, at roughly 25 percent probability, sees advantage recede as classical algorithms close the gap, fault tolerance prove harder to scale than 2024 to 2025 results suggest, and funding contract. IonQ, Rigetti, and similar listed names face strategic alternatives reviews, and the field consolidates around two or three sovereign-backed efforts. The upside case, at 15 percent probability, has a major modality deliver a usable few-thousand-logical-qubit machine before 2030, breaking materials simulation and chemistry open and triggering a step change in cryptographic urgency. Across all three states, the actionable variables for chief information officers and capital allocators in 2026 are the same: PQC migration plans, vendor benchmark transparency, and the cost trajectory of logical qubits.

Sources #

• IBM Quantum roadmap and processor announcements
• Google Quantum AI Willow announcement, December 2024
• NIST Post-Quantum Cryptography standards FIPS 203, 204, 205
• US National Quantum Initiative Office
• IonQ and Quantinuum investor and technical disclosures
• McKinsey Quantum Technology Monitor 2024 and 2025
• Boston Consulting Group quantum reports
• Nature, Quantum error correction below the surface code threshold (Willow)
• MIT Technology Review Quantum coverage
• IEEE Spectrum quantum computing reporting