The thesis on quantum computing moved materially in May 2026, not because a new qubit record was set, but because federal capital and public-market structure arrived simultaneously. The US Department of Commerce signed CHIPS Act letters of intent with both Quantinuum and D-Wave on the same day Quantinuum's IPO registration was already on file with regulators — a policy signal that the government has decided to backstop the domestic quantum supply chain on exactly the terms it backstopped advanced-node logic.
The government has picked its instrument
The 2026-05-21 dual LoI — D-Wave (QBTS) receiving a disclosed $100M commitment ([source: https://www.dwavequantum.com/company/newsroom/press-release/]), Quantinuum's amount undisclosed ([source: https://www.quantinuum.com/news]) — shares a news window with Quantinuum's registration statement, filed publicly on 2026-05-08 (confidential draft 2026-04-22). I read the timing as non-coincidental: the policy goal of stapling US private capital to US sovereign quantum capacity is now explicit, not inferred. DARPA is running the same play on the procurement side: IonQ (IONQ) was selected for the HARQ (Heterogeneous Architectures for Quantum) programme on 2026-04-14 ([source: https://ionq.com/news]), and the NSA's Florida LambdaRail quantum-safe network — also supplied by IonQ — was announced 2026-04-27 ([source: https://ionq.com/news]). Federal supply-side subsidy and defence demand are now underwritten in parallel.
The cryptographic clock is moving faster than the policy community realised
The most consequential research event in the prior twelve months is Craig Gidney's May 2025 update on Shor's algorithm resource requirements, which compressed the noisy-qubit overhead for factoring RSA-2048 from the 2019 estimate of 20 million qubits over eight hours to under one million qubits in under one week — a 20× reduction via approximate residue arithmetic, yoked surface codes, and magic-state cultivation ([source: https://arxiv.org/abs/2505.15917]). These are peer-reviewable algorithmic optimisations, not hardware speculation. The gap to demonstrated hardware remains four to five orders of magnitude: Quantinuum's H2 has shown fewer than 20 logical qubits; Google's Willow — 105 physical qubits, below-threshold error correction, December 2024 ([source: https://blog.google/technology/research/google-willow-quantum-chip/]) — demonstrates one logical qubit at a useful error rate. The cryptographic-relevance horizon therefore remains 8–15 years on Gidney's revised arithmetic, but the direction of revision is the signal: the bull case for breaking RSA is moving forward, not backward.
The policy layer has already priced this asymmetry. NIST finalised FIPS 203/204/205 post-quantum cryptography standards in August 2024 ([source: https://csrc.nist.gov/projects/post-quantum-cryptography]); NIST IR 8547 sets 2035 as the federal deprecation deadline for classical RSA/ECC; NSA's CNSA 2.0 mandates full transition by 2033. The result is a seven-year mandated procurement cycle now in force, running regardless of whether a fault-tolerant machine exists in 2030 or 2035. Cloudflare (NET) sits at the deepest publicly measurable point of that cycle by TLS handshake share; Akamai (AKAM), Thales (HO.PA), and the federal IDIQ contractors sit on the same demand curve and are worth tracking separately from the hardware pure-plays.
Where the durable economics sit
I map the value chain in six layers and find the most enduring economics at layer two: trusted-foundry mature-node control electronics. IonQ's $1.8B acquisition of SkyWater Technology, closed 2026-01-26 ([source: https://ionq.com/news]), is the clearest valuation signal in the theme. SkyWater runs 90nm and 130nm processes — well off the leading edge — but it is DoD Trusted Foundry-certified and ITAR-clean, making it near-uniquely valuable for trapped-ion control electronics that require radiation-hardened, supply-secure mature-node capacity rather than Moore's-law advancement. IonQ paid for sovereignty. Combined with the Oxford Ionics acquisition (~$1B, September 2025), IonQ has deployed roughly $2.8B on M&A in seven months — a capital intensity that strains the "venture-stage public company" framing considerably. GlobalFoundries (GFS) occupies the equivalent trusted-foundry position for photonic quantum via its Malta NY fab, which runs PsiQuantum's wafer process.
Layer one — cryogenics, lasers, and vacuum systems (Bluefors, Oxford Instruments (OXIG.L), Coherent (COHR)) — is similarly durable and sells into every modality regardless of which architecture wins. Layer three, the processor pure-plays, is the most exposed to commoditisation: a five-modality race (superconducting, trapped-ion, photonic, annealing, neutral-atom) is structurally hostile to durable economics until consolidation occurs. IonQ and Rigetti (RGTI) carry option value on trapped-ion and superconducting respectively; IBM (IBM) provides the picks-and-shovels narrative inside the hyperscaler set alongside Google (GOOGL), running the Loon → Kookaburra → Cockatoo → Starling (2029 target: 200 logical qubits, ~100M gate operations via bivariate-bicycle qLDPC codes) → Blue Jay sequence. End-buyer adoption is beginning to verify: Synopsys (SNPS) via a Quantinuum partnership dated 2026-05-19, bp (BP) for wave-physics modelling on 2026-05-20, and BMW Group via a Quantinuum partnership on 2026-05-05 are the first named industrials with disclosed engagements.
The Quantinuum IPO will reset the comp set
Quantinuum's pending IPO is the highest-impact capital-markets event in the theme this year. If it prices in the $8–15B range — inferred from 2024 funding-round comparables — it will either re-rate IonQ (~$11B market cap on ~$50M trailing revenue) and QBTS upward as peers, or pull capital toward the higher-quality issuer. Once Quantinuum enters major US indices, passive flow becomes a meaningful component of the marginal price-setter on a name with revenue measured in tens of millions — the same mechanism already compressing fundamental analysis in IONQ and QBTS today.
What would change my mind
The Quantinuum IPO postponing past 2026 or pricing below a $6B implied valuation would falsify the "federal capital and public-market structure arriving simultaneously" thesis and suggest institutional underwriting appetite is shallower than the LoI timing implies. IonQ reporting Q3 2026 revenue below $20M sequential would signal the SkyWater integration is consuming organic execution bandwidth rather than expanding it. A credible demonstration of more than 100 logical qubits at a fault-tolerant error rate from a Chinese laboratory — Pan Jianwei's USTC group or Origin Quantum — would materially strain the geopolitical-decoupling rationale underpinning the CHIPS Act subsidy logic. Any one of these would require a full reassessment.
What I'm watching next
Three leading indicators in order of nearness: the Quantinuum IPO pricing and its implied comparable for the IONQ/RGTI/QBTS multiple set; IBM's delivery against the Kookaburra 2026 milestone on its public roadmap (primary IBM source URLs returned 404 this scan — verification is queued for the next cycle); and the close of CHIPS Act funding tranches on the Quantinuum and D-Wave LoIs by Q4. Further out, the 2027–2028 window is the architectural-race resolution test: I want at least one published industrial chemistry result demonstrating quantum advantage on a paying-customer problem, and consolidated quantum cloud revenue across AWS Braket (AMZN), Azure Quantum (MSFT), IBM Quantum, and Google Quantum AI crossing $1B per year — the clearest proxy for whether enterprise demand is real or still pre-commercial.