D-Wave Quantum Inc. QBTS

Key data

The business

D-Wave sells quantum computing two ways: the physical Advantage2 machine placed on-premises, and Leap, a cloud subscription where customers rent time to run optimization problems, the kind of scheduling, routing and portfolio math that grows ugly fast on ordinary computers. It is the annealing specialist, a narrower and older design than the gate-model machines IBM, Google and IonQ are building, and it has now added its own gate-model line. The pitch is that annealing is useful for real commercial work today, not a decade out. More than 100 organizations across commercial, government and research sectors touch it. The moat, such as it is, is a fifteen-year head start in annealing hardware and the switching friction once a customer builds a workflow on Leap. The problem is that almost nobody has yet decided quantum is worth paying real money for.

The numbers

Two things matter here, and the headline hides both. Start with the quarters.

The enormous net losses in mid-2025 are mostly non-cash, a re-marking of warrant and earnout liabilities as the stock ran toward $47. Operating loss those quarters was closer to $27 million. Ignore the accounting noise and read the top line: revenue has sat near $3 million a quarter for over a year, flat, with no visible ramp.

That 2025 jump to $24.6 million is the whole bull story, and it is one sale. A single Advantage2 system booked $15.0 million in Q1 2025. Strip it and the rest of 2025's revenue was about $9.6 million, barely ahead of 2024's $8.8 million. The run-rate business is roughly $12 million a year and barely grew. Against that, consensus asks for $148 million by 2028, a six-fold climb in three years, and still a loss at the bottom. The market is paying about 500 times trailing sales for a revenue line that, ex the one machine, barely grew. What I believe that the tape does not: the 2025 print was an event, not an inflection, and the only number that decides this company is recurring Leap revenue, which runs near $12 million annualized. The print that settles it is QCaaS bookings and Customer A's share in the next two 10-Qs. Liquidity is not the near-term worry: working capital is about $533 million, several years of runway even at an elevated burn, funded by selling stock.

Management

Follow the cash, and it runs one direction. Insiders sold about $112.6 million across 46 transactions in the last year against a single $1,795 purchase. CEO Alan Baratz led it, roughly $65 million across three sales, $23.9 million in December 2025, $23.3 million in November 2025, $18.0 million in June 2026. Plan status is not disclosed in the Form 4 footnotes, so I will not call these routine. Meanwhile the company funds itself by issuing shares, and stock-based compensation runs about 250% of revenue. The pattern is plain: management is converting equity to cash at scale while the public funds the burn. That is a governance flag, not a fraud claim, but on a pre-revenue-scale name it is the loudest signal on the page.

How it fails or surprises you

Customer concentration. Customer A was 41% of revenue in Q2 2026. Lose that account and revenue roughly halves overnight, with no diversified base to cushion it. Watch the significant-customer table in the next two 10-Qs; a share climbing past 50% is the warning, a departure is the break.

A real commercial win (right tail). If a marquee enterprise puts an optimization workload into production and it scales into seven-figure recurring spend, the durability the market prices near zero gets repriced fast. Nobody is paying for this at scale today. The print is a named multi-year Leap contract or a step-change in QCaaS bookings inside the next four quarters.

Repeatable system sales. My read assumes the $15 million Q1 2025 machine was a one-off. If D-Wave books a second large system in 2026 or 2027, and then a third, the "recurring is flat" thesis is simply wrong and the revenue base is lumpier but real. A second big system sale would prove me wrong first.

Closing thoughts

Nothing in the next four quarters resolves whether quantum annealing becomes a paying business; the timeline is measured in years, and the balance sheet buys those years. So the question is not survival, it is credibility: does revenue inflect before the market's patience, and the share count, run out. Near term the left tail is fatter. A stock already down 65% from its high, one customer at 41%, insiders selling into every rally, and a valuation at 490 times trailing sales all lean one way, and the right tail, though genuine, is a bet on timing the market will not pay for until it sees bookings.

The bet is still that companies keep paying D-Wave to solve hard scheduling and logistics problems with its quantum machines, and that a handful of paying customers turns into hundreds. It breaks if revenue stays near $12 million annualized and Customer A walks. The pair of numbers that tells you first, before any headline: quarterly revenue, and Customer A's share of it. Until both move together, the machine sale is a story, not a trend.