Quantum computing

Core thesis

Quantum computing has made genuine technical progress through 2025-2026 — Google's Quantum Echoes algorithm on the 105-qubit Willow chip demonstrated the first verifiable quantum advantage (13,000x speedup, October 2025), IBM shipped its 120-qubit Nighthawk processor (November 2025) on a path to the 2029 Starling fault-tolerant machine, and IonQ became the first pure-play quantum company to surpass $100M annual revenue ($130M for FY2025, up 202% YoY). But commercial utility for real-world problems remains 5+ years away.

The key technical milestone is fault-tolerant quantum computing with logical qubits. The current record is 96 logical qubits (QuEra, Harvard/MIT/Caltech, Nature, Jan 2026), with Quantinuum's Helios running 48 fully error-corrected logical qubits. The threshold for practical drug discovery or cryptography runs from ~1,000 logical qubits into the thousands, with error rates well below threshold. No vendor has a credible path to that scale before the end of the decade.

The most likely first commercial applications are: (1) pharmaceutical molecule simulation, where quantum can model molecular interactions intractable for classical computers; (2) materials science, particularly catalyst design and battery chemistry; (3) financial portfolio optimisation under complex constraints. General-purpose quantum advantage in ML or database search remains theoretical.

State of the art (2026)

As of mid-2026 the hardware race is delivering, not stalling. Quantinuum's Helios (launched November 2025) runs 98 trapped-ion qubits at sector-leading fidelity; IBM unveiled its 120-qubit Nighthawk processor and reaffirmed a 2029 target for Starling, a 200-logical-qubit fault-tolerant machine; and Google's Quantum Echoes result on the 105-qubit Willow chip (October 2025) marked the first verifiable quantum advantage, a 13,000x speedup. IonQ's $1.8B SkyWater acquisition, approved by stockholders in May 2026, signals a vertical-integration land grab. Yet logical-qubit counts remain in the tens, error-corrected utility for chemistry or cryptography is still 5+ years out, and the field has no consensus winning architecture. The near-term money is in post-quantum cryptography migration, not computation.

The hardware race is fragmenting, not converging

Five competing hardware approaches are all making progress, which is unusual this late in a technology cycle. Superconducting (IBM, Google), trapped ion (IonQ, Quantinuum), neutral atom (Atom Computing, QuEra, PASQAL), photonic (PsiQuantum, Xanadu), and topological (Microsoft) each have different strengths. Superconducting leads on gate speed; trapped ion leads on fidelity; neutral atom leads on qubit count; photonic promises room-temperature operation.

This fragmentation means the industry hasn't found its 'x86 moment' — the architecture that wins and becomes the standard. Until it does, enterprise customers face a platform risk that deters large-scale investment. The consolidation catalyst will likely be whichever approach first demonstrates fault tolerance at scale.

The real near-term opportunity is post-quantum cryptography, not quantum computing

While quantum computing is 5+ years from utility, quantum threats to cryptography are already driving spending. Harvest-now-decrypt-later attacks mean sensitive data encrypted today with RSA/ECC is vulnerable to future quantum computers. NIST finalised post-quantum cryptography standards (FIPS 203/204/205) in 2024, and the migration deadline for US federal systems is 2030. This PQC migration — estimated at $10-50B across government and enterprise — is the largest near-term revenue opportunity in the quantum ecosystem, and it doesn't require quantum computers to exist.