Open in interactive viewer → charts, metric popovers & call review

QCI Sold $9.5 Million of Photonics and Kept 4.5% of It

Record revenue, a $1.3 billion treasury doing most of the earning, and a Dirac-3 benchmark management refuses to call supremacy
QUBT · Earnings Call · 2026-11-09

A Record Quarter With a Hole in the Middle

Quantum Computing Inc. reported its third quarter on November 9 with the loudest top line in its history: $9.5 million of revenue against $384 thousand a year earlier, a 25-fold step-up driven by commercial foundation businesses acquired this year — photonic products sold into aerospace, government and industrial accounts. The number is real. It also beat the company's own first-half total, which management was keen to flag. What management was less keen to flag is what it cost to produce. “We generated $428 thousand in gross profit during the quarter, for a gross margin of 4.5%. While margins remain modest, we are encouraged by the progress.” — Christopher Bruce Roberts · 2026-11-09 Modest is one word. On total revenue, QCI has now booked roughly $10 million in the first half of 2026 plus $9.5 million in Q3; but gross profit spent the first two quarters of 2026 below zero (about -$1 million each) and only crossed into positive territory in Q3. A gross margin in the low single digits is what an underutilized fab looks like — and QCI now operates eleven facilities across eight states, which is precisely why the CEO spent a chunk of the call narrating parts moving between sites: “we fabricate some parts in our Santa Barbara site and some parts, in our in our Boston site, and so then we bring them to our prison site, and then the prison site is actually working with our North Carolina site” — Yuping Huang · 2026-11-09 (the transcript's own keyword extraction christened that prison site — it is Princeton, but the sprawl is the point). This matters because the company told investors in May that margins would normalize with volume. “We should be able to get back to 20% to 30% as volume picks up” — Christopher Roberts, CFO · 2026-05-11 — that was the CFO in the prior-year Q&A. Volume picked up twenty-five times and the margin did not. Instead the quarter's profit came from somewhere else entirely: $13 million of interest and other income against $428 thousand of gross profit, on $1.3 billion of cash and investments, producing a $15.1 million net loss despite it. Strip the treasury yield out and this is a company spending $29 million a quarter to earn 4.5% on its product. At 223.5 times revenue, the market is still paying for a future, not this present.

A Benchmark, Carefully Worded

The genuinely new event of the quarter is the Dirac-3, a room-temperature photonic optimizer scaled to roughly 9.98 thousand variables, with first customer shipment expected in Q4. What is new isn't the machine — it is the comparison set. This quarter, the company's argument is framed against classical solvers, which is the single highest-ranked theme in QCI's own keyword set for the period and was not meaningfully present a year ago. Yuping Huang named Hexaly and literature benchmarks; he also conceded “there are, of course, some problems that our machines either comparable or less optimal in terms of getting the solutions.” — Yuping Huang · 2026-11-09 Analyst Nehal Chokshi had to press twice to get the solver named at all, with the CFO audibly fumbling the handoff. And then the tell:

I do not want to make a broad claim that we now have achieved the quantum supremacy, and it is not a good word. Anyway, but I think as of now, with Dirac-3, there is certainly a quantum advantages that customers can access today.

Yuping Huang · 2026-11-09
Note the shift in opponent. In August, the same CEO described benchmarking “against other optimization machines using quantum effects” — Yuping Huang, CEO · 2026-08-10. Now the yardstick is the classical incumbent — a more commercially honest framing, and also an admission that the quantum-vs-quantum scoreboard never mattered to buyers.

The Threads That Went Quiet

The most interesting thing about this call is what is missing from it. QCI's first-quarter acquisitions included NuCrypt, a quantum-communications asset the CEO once pitched as potentially

one of the very first Quantum technology that can be adopted by large population

Yuping Huang, Chief Executive Officer · 2026-03-02
— a network-security product with a bank reference customer. On this call, communications got one subordinate clause: the platform extends beyond computing into sensing and communications as well. NuCrypt is not named once. In QCI's own keyword book, NuCrypt was the company's hardest-falling theme of the quarter, and room temperature — the physical differentiator that anchored two years of investor decks — declined alongside it. The gate-based machine, which in March occupied a multi-sentence answer about five extreme optical conditions and resonator quality factors of two million versus a ten-million target, also faded to a single line about integrating more technology onto chips. What replaced them is a portfolio of nearer-term stories: an trial study collaboration with ImmuBio in clinical data (the first pharma customer), drug discovery adjacency, and a promise of heterogeneous integrated circuits within 12 to 18 months. The recurring phrase scalable manufacturing is not new either — it has been in QCI's top keyword set since the first quarter. So when the CEO frames this as a new stage, “QSI is really transitioning from a tech innovation company to a scalable manufacturing con. Company.” — Yuping Huang · 2026-11-09 — recall that the identical transition was announced in May. The language of novelty is two quarters old.

The Pivot Meets the Tape

The freshest strategic thread is a detour into data centers, which jumped to QCI's second-ranked theme on the back of one answer. Asked about the political environment, the CFO volunteered that AI and data centers are the consensus target and that “we have done some analysis on our own looking at the application of Dirac optimization to improving the efficiency of the utilization of compute resources in data centers” — Christopher Bruce Roberts · 2026-11-09 — squeezing more out of existing racks rather than building new ones. It is a genuinely company-unique angle on a very crowded theme; the recent-reporter tape is wall-to-wall compute-buildout language from far larger companies. The problem is the price tape. Over the last 90 days the market has been actively de-rating exactly that complex: data-center capacity names down roughly 35%, AI data-center exposure negative across dozens of tickers, high-performance computing down nearly 30%. QCI is pointing its narrative at the one pocket of the market currently being sold. Its own stock has bled 26.8% over the same ninety days into this print, sitting at its drawdown from a July high near $9.37, on a $1.43 billion market cap — no longer a squeeze story, just a slow re-rating. What to watch is small and specific. The backlog slipped to $39.1 million from $42.5 million, which the CFO waved off as noise: “our backlog was down like $2 million from where we were” — Christopher Bruce Roberts · 2026-11-09 — though last quarter the same management stood by “the models that we have seen that showed QCI before enhanced doing $20 million to $25 million For this year, we stand by that” — Christopher Bruce Roberts, CFO · 2026-08-10, and that number has quietly stopped being repeated. Guidance remains absent by design. Roughly 70% of the business is still government subcontract work. And $70 million to $80 million of Fab 2 spending is scheduled to leave the building by the end of March. The cash buys time; the gross margin is what has to move.