Question: Is quantum computing "somewhat working" now — and is any of it investable for a trend-hold book?

Verdict: Somewhat. The science crossed a real, verifiable advantage line in late-2025; commercially, only D-Wave (QBTS) has an actual business, and even that is small, lumpy, and in a narrow niche. Everything else is a capital-cycle / balance-sheet-survivor story — companies that raised enormous cash into the 2025-26 quantum mania, not businesses that earn it. The trade, if there is one, is survivors over lottery tickets, fade the pumps, no buy today (the whole basket is in a drawdown and insider-sold).

This is the coverage synthesis for the quantum-compute deep-dive program, built on five primary-source 10-K/10-Q deep-dives dated 2026-06-28 (INFQ, IONQ, RGTI, QBTS, QUBT), the existing LAES deep-dive (2026-06-25), and an EDGAR pull on ARQQ.

What we're asking

"Is quantum even somewhat working now?" has two different answers depending on which "working" you mean:

1. Does the technology do anything real yet? (the science layer)
2. Is any of it a business — and is any of it a trade for our trend-hold book? (the commercial + tape layer)

Conflating those two is how people lose money in this group: the science genuinely improved, so the stocks ran, but the businesses didn't arrive — and the cash on the balance sheets came from issuing shares into the narrative, not from selling quantum computing.

What we found

Layer 1 — the science: yes, it crossed a real line in late-2025

These are not vaporware milestones:

• Google Willow "Quantum Echoes" (Oct 2025): a verifiable quantum-advantage result — ~13,000× a top supercomputer on an OTOC physics problem (Google Quantum AI, Oct 2025; blog.google / Nature) — with a proof-of-concept NMR/molecule application (15- and 28-atom molecules) pointing at drug-discovery / materials use. This is the strongest "it works" data point. But it's Google (not an investable pure-play), and it's a science milestone, not commercial revenue.
• D-Wave (QBTS): peer-reviewed "Beyond-Classical Computation in Quantum Simulation" — claimed first useful-problem quantum supremacy (materials sim). Annealing, not gate-model — a narrower result than Willow, but on a useful problem and from an investable company.
• IonQ (IONQ): first to cross 99.99% two-qubit gate fidelity ("four nines") in R&D prototypes; concrete #AQ-denominated roadmap.
• Rigetti (RGTI): shipped a 108-qubit general-availability system (Cepheus-1-108Q, Apr 2026) you can actually run jobs on, plus real-time error correction with Riverlane.

So: at the physics/benchmark layer, "somewhat working" is YES. Fidelity, qubit counts, error correction, and a genuine verifiable-advantage result all moved in 2025-26.

Layer 2 — the business: mostly no, with one real exception

Primary-source scorecard (FY2025 unless noted; price/tape from the scan summaries, Fri 2026-06-26 settled close; financials code-pulled from SEC EDGAR XBRL; liquidity = cash & equivalents + short-term investments). Per-name price-truth and figures are cited in each name's own deep dive:

*INFQ insider "−$538M" is de-SPAC sponsor/PIPE mechanics, not distribution — see its deep-dive. ARQQ/LAES are PQC-adjacent foreign (20-F) filers, a different category (LAES already has a full deep-dive, 2026-06-25; ARQQ's is deferred) — see scope note below.

Three things fall out of that table:

1. Only one name has a real commercial business: D-Wave (QBTS). Record Q1'26 bookings of $33.4M (+~2,000% YoY) — a $20M Advantage2 system sale (Florida Atlantic) + a $10M Fortune-100 QCaaS deal — 100+ customers (>50% commercial), 100+ orgs in production, usage +314% YoY. That is a business. The catch: it's annealing (optimization/sampling), a narrower niche than the general-purpose machine the sector's valuations price.

2. The "revenue" that exists elsewhere is mostly government/contracts or a bolt-on. IONQ's $130M (the most) is heavily government/contract-weighted and decelerating; RGTI's $7M leans on DARPA/AFRL awards ("growth model too dependent on government deals"); INFQ's $32.5M is its sensing hardware business (atomic clocks/RF), not quantum computing; QUBT is effectively pre-revenue ($0.68M); ARQQ is ~zero.

3. What's actually "working" spectacularly is fundraising, not the P&L. IONQ ~$2.0B, QUBT ~$986M, QBTS ~$588M, INFQ ~$569M, LAES ~$428M of liquidity — raised by issuing stock into the mania. The income statements are oceans of red (IONQ −$634M op loss; ~$280M/yr cash burn). And every "profit" quarter in the group (IONQ +$805M, RGTI +$33M net) is a non-cash warrant/derivative mark-to-market swing, not operating earnings — a recurring trap flagged in each deep-dive.

Layer 3 — the tape: the basket is in a drawdown, insiders are selling

Every name is down over the trailing 30 sessions except ARQQ (+47% momentum pop on near-zero revenue). And insiders are net sellers across the entire group (IONQ 55 sells, QBTS 56, RGTI 50, QUBT 12, LAES 20, ARQQ 8 — vs a literal handful of buys). The people closest to the technology are distributing into retail momentum. That is the textbook signature of the channel's own framing — "social arb, fade the pumps, real winners 1-2yr out" — and it matches our house skeptic view.

Verdict + reasoning

Is quantum somewhat working? Yes at the physics layer, barely at the business layer, and not at all as a trend-hold entry today.

• Science: working. Verifiable advantage (Willow), a useful-problem supremacy result (D-Wave), four-nines fidelity (IonQ), a 108-qubit GA system (Rigetti). Real, in the last ~9 months.
• Business: one real exception. D-Wave has actual bookings/customers/production workloads — in a narrow annealing niche. Everyone else is pre-commercial, government-funded, or (QUBT/ARQQ) essentially a balance sheet with a science project attached.
• Trade: survivors over lottery tickets, no buy yet. If quantum graduates from "science working" to "business working," the path runs through the revenue-and-balance-sheet survivors — QBTS (real commercial traction), IONQ (most revenue + fortress balance sheet + fault-tolerant roadmap), and INFQ (real sensing revenue funding the bet) — NOT the pre-revenue tickets (QUBT, RGTI on dilution risk, ARQQ). But none is a trend-hold entry today: all loss-making, all insider-sold, the whole basket in a drawdown. Watch the survivors for a base
  • a clean post-deal-cost / bookings-converting quarter; fade the pumps.

Scope note — ARQQ & LAES: these are post-quantum-cryptography / secure-chip names (ARQQ = Arqit PQC; LAES = SEALSQ secure semiconductors), foreign 20-F filers, a different category from gate-model/annealing quantum computing. LAES already has a full deep-dive (2026-06-25-laes-deep-dive.md); ARQQ is scorecard-only here (full deep-dive deferred). Both belong to a separate PQC sub-thread, not the gate-model/annealing quantum-computing program. (Consistent with ARQQ's deliberate exclusion from the quantum-pure-play thesis in quantum-computing watchlist notes.)

Program status: this completes 5 of the 7 quantum-computing-universe deep-dives (INFQ + IONQ + RGTI + QBTS + QUBT). The program's graduation rule ("≥3 deep-dives land and a thesis converges → graduate to a live perspective, demoting a monitoring one for the 8-cap") is now met on the count and the thesis does converge — namely "quantum is a capital-cycle / balance-sheet-survivor trade, not a commercial-advantage trade; own survivors, fade pumps, no buy until a base." Recommend the user decide whether to graduate a quantum-compute-capital-cycle perspective (requires demoting a monitoring perspective for the 8-cap — a user-managed call, not auto-done here). Until then the house view stays skeptical-but-watching and no perspective is opened.

Sources

• Deep-dives (2026-06-28): 2026-06-28-{infq,ionq,rgti,qbts,qubt}-deep-dive.md; plus the existing 2026-06-25-laes-deep-dive.md (PQC sub-thread).
• Program: TASKS-RESEARCH — quantum-compute deep-dive program (user-directed 2026-06-25; source note 2026-06-27).
• Watchlist: quantum-computing (IONQ/RGTI/QBTS/QUBT/INFQ/ARQQ/LAES). None held.
• Capture origin: 2026-06-27-dumb-money-discord-3channel-sweep (⚛️ Quantum) — social surfaced the names; this program verified them against filings.
• Cross-link (not ownership): QUBT's TFLN photonic foundry touches the optical-supercycle perspective (datacom/telecom modulators); cited as evidence, not a placement.