Rocket Doctor's Q2 Wasn't Flat. It Was Deferred.

Disclosure: I am long Rocket Doctor AI Inc. (CSE: AIDR | OTC: AIRDF | FSE: 939). I hold over 1.3 million shares, which I accumulated at various prices. I am not compensated by the company or by anyone connected to it. This is not investment advice. Do your own work.

On September 1, Rocket Doctor AI filed its Q2 2026 interim report. Revenue came in at $734,028, against $737,103 in Q1. Flat. Down $ 3,000 quarter over quarter.

In the same quarter, completed U.S. patient visits went from 1,319 to 3,911. A 196% increase.

Those two facts do not belong in the same quarter unless something structural is separating operations from the income statement. Something is. It’s in the revenue recognition policy; it’s disclosed in plain language in the filed MD&A, and almost nobody reading the headline number appears to have registered it.

What the policy actually says

From the filed MD&A (this is the Q1 language, carried into Q2):

For its U.S. operations, the Company recognizes revenue on a cash basis as it works through the complexities of billing Medicare, Medicaid, and private insurance providers for patient visits... the ultimate receipt of payment is not reasonably assured at the time services are delivered. As a result, the Company has determined that collectability cannot be established with sufficient certainty to satisfy the revenue recognition criteria required under IFRS 15.

This applies the variable consideration constraint in IFRS 15 at its most conservative setting. Under a normal accrual policy, you recognize revenue when the service is delivered and book a receivable, net of an expected credit loss allowance. Rocket Doctor doesn’t do that in the U.S. It waits for the cheque.

Management’s stated rationale is that it has no collection history to build an estimate from. That’s a legitimate position: a new entrant billing Medicaid across three states genuinely cannot model realization rates from nothing. It is also, mechanically, a policy that makes a fast-growing business look stationary.

Why the gap widens when you grow

Here is the part that matters, and it’s not intuitive.

Under cash-basis recognition with a fixed collection lag, the revenue you report this quarter is a function of the visits you delivered last quarter (or the one before). If visit volume is flat, that’s harmless; the numbers converge. If visit volume is compounding, the reported figure is permanently anchored to a smaller past.

Q2 recognized cash from a period when the company was running roughly 1,300 visits a quarter. It delivered 3,911. The tripling doesn’t show up. And if Q3 triples again, Q3’s reported revenue will reflect Q2’s 3,911, still behind.

The lag isn’t the problem. The acceleration is what makes the lag expensive. The faster the business grows, the more understated it looks, and the understatement compounds rather than closes.

The arithmetic

The US$25-per-visit figure gets quoted constantly and almost always misused. Maxim Group’s initiation model is US$18. Neither number is a revenue figure.

Per management commentary, Dr. Essam Hamza and Dr. Bill Cherniak have both described the model: a U.S. visit is recognized at roughly US$125 gross, with roughly US$100 paid out to the treating physician as a direct cost. The company keeps approximately US$25. That's where the US$25 figure comes from, and it is a net take, not a top line.

This detail reframes everything, and it is the one most people get backwards. Two separate numbers fall out of the same visit:

• Recognized revenue per visit: ~US$125 (~C$169)
• Gross profit per visit: ~US$25 (~C$34)

(Converted at approximately 1.35 CAD/USD throughout. The US$125 / US$100 split is management commentary from investor communications, not a figure disclosed in the filed financial statements; treat it accordingly.)

The filed Q1 segment note corroborates it. Note 12 of the condensed interim statements reports U.S. revenue of C$179,350 for Q1 2026 on 1,319 completed visits, roughly C$136 per delivered visit. Against a C$169 gross recognition rate, that implies about 80% of Q1’s delivered visits had converted to cash by quarter-end. The number that looked anomalous against a US$25 assumption is exactly what you’d expect under gross recognition with a collection lag.

Now run Q2.

I don’t have the Q2 segment split yet, so I derived it. Q2 gross margin was 65% on $734,028 of revenue, implying direct costs of roughly C$257,000. Using Q1’s filed figures to back out the Canadian gross margin, C$557,753 of Canadian revenue against Q1’s C$184,529 of direct costs, once you allocate 80% of U.S. revenue to physician cost, Canada runs at roughly 92-93%. Solving the Q2 blend for those two margins gives:

• Q2 U.S. revenue: roughly C$279,000 (up ~56% from C$179,350 in Q1)
• Q2 Canadian revenue: roughly C$455,000

At C$169 per visit, C$279,000 of recognized U.S. revenue represents about 1,650 visits’ worth of collections. The company delivered 3,911.

That leaves roughly 2,260 delivered visits with no recognized revenue, aboutC$382,000 of revenue sitting in the billing pipeline rather than the income statement.

Accrue it, and Q2 revenue is roughly C$1,116,000 rather than $734,028. Not flat. Up about 52% sequentially.

And here is the part that should change how you read the margin line. That C$382,000 of deferred revenue carries only about C$77,000 of gross profit, because the physician cost travels with it. Accruing it fully would push Q2 blended gross margin down to roughly 50%, not up.

Which means the four-quarter margin compression everyone is flagging — 89% to 75% to 65%, is not deterioration. It is arithmetic. A business mixing from a 92% Canadian line into a 20% U.S. gross-recognition line will see blended margin fall toward 20% as it succeeds. The margin chart going down is the same event as the U.S. business going up. Those are not two facts; they are one fact viewed twice.

To be clear about what is mine and what is filed: the C$279,000 U.S. revenue figure and everything downstream of it are my derivations from the disclosed 65% gross margin and Q1’s filed segment note. They are not company disclosures. The Q2 segment note on SEDAR+ will confirm or refute them, and I’d rather be corrected publicly than quoted without the caveat.

Three things this argument does not survive if ignored

Out-of-network visits. The filed MD&A is explicit: the company deliberately permits out-of-network patients onto the platform to build volume, and reimbursement for those visits is generally not expected. Claims are submitted anyway, in case a payer approves them. We don’t know what share of the 3,911 were out-of-network. If it’s a third, a third of the accrual estimate evaporates. This is the first thing a competent bear will raise, and it deserves a straight answer from management.

Revenue is understated far more than earnings are. C$382,000 of deferred revenue carries roughly C$77,000 of gross profit. If you’re going to quote the top-line number, quote the profit number beside it. Anyone selling the deferral as a hidden earnings pool is selling something, and the arithmetic is right there to catch them.

None of it fixes the burn, but it helps you read the burn correctly. The headline $6.67 million net loss is not a cash number. A large share of it is non-cash: depreciation and amortization, share-based compensation, the change in fair value of contingent consideration, and the marketing cost of the Rick Ware Racing / FinTekk partnership, which is settled in stock at $0.70 rather than cash. So roughly half of that loss was a non-cash item related to the Rick Ware/ FinTekk partnership. Quoting $6.67 million as if the company burned $6.67 million is wrong, and it’s the single most common error in the bear commentary on this name.

The cash flow statement is the honest place to look. Cash went from C$3,495,589 at March 31 to C$0.95 million at June 30, roughly C$2.55 million consumed over the quarter, on the order of C$850,000 a month. That is a materially different picture from the headline loss, and it is the number I use.

And non-cash is not free. Paying FinTekk in stock at $0.70 is dilution, not a discount. It belongs on the share count, not in the “ignore this” column.

At C$850,000 a month, the C$0.95 million on hand at June 30 was about five weeks of runway. That is why the C$3.26 million of convertible debentures got raised in August, and those add another layer to a dilution stack that already includes the escrow release schedule and the January 2027 warrants. The recognition argument is about whether the operating trajectory is visible in the financials. It is not a solvency argument, and it should never be used as one.

What I’m actually watching

The thesis has never rested on covered lives. Twenty-four million in-network lives describes access, not utilization, and I’ve said so every time I’ve written about this company.

Q2 gives us the first quarter where the credentialing constraint visibly loosened: active U.S. clinicians went from 19 to 30, with several onboarded mid-quarter, while 80 total on the roster were still moving through credentialing. Visits nearly tripled off that. Per-physician throughput, not headline visit count, is the metric that tells you whether this scales.

The specific things I want out of Q3:

• The U.S. segment revenue line, to test my ~C$279,000 Q2 derivation and to separate patient billing from NIH/service revenue
• Whether the company discloses in-network versus out-of-network visit mix
• Implied collection rate: U.S. revenue divided by C$169, against visits delivered. Q1 ran near 80%; my Q2 derivation puts it near 42%. That drop is what a step-change in volume looks like, but it is also what a collections problem looks like
• Cash used in operating activities, to confirm whether the ~C$850,000 monthly burn is real or flattered by stretched payables
• Blended gross margin continuing down toward the 20% U.S. floor, which I would read as confirmation rather than deterioration
• Any move to reassess the recognition policy once collection history exists, which management has said it will evaluate

If Q3 revenue is flat again while visits climb again, one of two things is true: either the lag is exactly what management says it is and the catch-up is coming, or the claims aren’t getting paid. Those look identical for one quarter. They do not look identical for three.

I think the first is more likely. I’m positioned accordingly, and I’ve told you the size of my position so you can weigh this appropriately.

Sources: Q2 2026 results press release (GlobeNewswire, September 1, 2026); Q1 2026 condensed interim consolidated financial statements and MD&A as filed on SEDAR+; Maxim Group initiation of coverage. Maxim Group makes a market in the Company’s securities. Fundamental Research Corp. coverage of the Company is paid coverage. Per-visit figures attributed to management are drawn from investor presentations and management commentary, not from filed financial statements.

Disclosure, again: I am long AIDR over 1.3 million shares. I may buy or sell at any time without notice. Nothing here is investment advice. Not financial advice. Do your own research.