CMPS — Q2 2026 Financial Report Analysis

Spending shifts from trials to launch prep; a $206 million paper loss swamps the quarter

COMPASS Pathways is a London-based drug developer with no product on the market yet. Its only late-stage candidate, COMP360, is a synthetic form of psilocybin (the active compound in "magic mushrooms") given in a supervised session for treatment-resistant depression (TRD): depression that hasn't improved after at least two standard antidepressants. Because the company has no sales, the useful questions are how fast it spends money, what it spends it on, and how long its cash lasts. It reports in US dollars and files the same 10-Q quarterly reports as a US company.

In the second quarter of 2026 (April to June), COMPASS reported a net loss of $253.8 million, up from $38.4 million a year earlier. Nearly all of that jump, $205.6 million, is an accounting charge on warrants that costs no cash (explained below). The figure that reflects the business is the operating loss (what it cost to run the company before financing items): $52.4 million, up 22%. Behind it, spending moved away from research and toward getting ready to sell the drug. General and administrative costs rose 84% to $23.2 million for commercial preparation, while research and development fell 4% to $29.2 million as the two big Phase 3 trials wind down.

At a glance

• $433.3 million in cash at June 30, up from $149.6 million at the end of 2025 after about $370 million of fundraising in the first half. Management says this funds the company "into 2028", past the expected launch.
• $52.4 million operating loss, up 22% from a year ago. Most of the rise is launch preparation: G&A went from $12.6 million to $23.2 million.
• $205.6 million non-cash warrant loss. This is why the net loss is six times last year's. The share price went up, which makes a liability on the balance sheet bigger. No cash leaves the company.

The numbers

For the first half of 2026, the net loss was $162.6 million ($1.33 per share), compared with $56.3 million ($0.62) in the first half of 2025.

Where the money went

Research and development: lower trial costs. R&D fell $1.2 million from a year earlier. Within it, "development expenses" (mostly paying for clinical trials) dropped from $21.4 million to $19.2 million. The 10-Q says this is "primarily due to lower clinical trial expenses as our Phase 3 program for COMP360 psilocybin therapy in TRD progresses toward completion," plus lower discovery-research costs after programs were cut in a late-2024 reorganization. External consultants and contractors pushed facilities and other R&D costs up from $1.5 million to $2.2 million, which partly offset the decline. The company does not expect R&D to keep falling. It says costs will stay "significant" until both the TRD program and the new PTSD trial are complete.

General and administrative: building a commercial team. G&A rose $10.6 million. Each line in the breakdown went up:

• Personnel: $8.2 million vs $4.6 million, from "increasing staffing levels to support our commercial preparedness activities."
• Facilities and other: $6.3 million vs $2.3 million, from consultants, contractors, and technology and software for launch.
• Legal and professional fees: $5.9 million vs $3.4 million, mostly consulting for commercial readiness.

This is the main signal in the quarter. A trial-stage company is turning into one that expects to sell a product. G&A was $16.4 million in the first quarter of 2026 (first-half $39.6 million minus Q2's $23.2 million), so it rose 41% from one quarter to the next. Expect it to keep rising before any revenue comes in.

Why the net loss is so large: warrants

A warrant is a contract that lets an investor buy shares at a set price later. COMPASS used warrants in its January 2025 and February 2026 fundraisings. Some of them are pre-funded warrants: the investor paid almost the whole price upfront and only has to pay $0.0001 per share to convert. Investors use them when they don't want to hold too many shares directly.

Under US accounting rules, these warrants are recorded as a liability because some contingent terms could in theory require a cash settlement. The company must revalue that liability every quarter at the current share price. When the share price rises, the liability grows and the increase is booked as a loss. By June 30, the warrant liability was $337.5 million, and the 10-Q says all of it was pre-funded warrants, since the 2025 ADS warrants had been fully exercised. Given the $0.0001 exercise price, these will almost certainly turn into shares, not cash payments.

The charge also swings in both directions. The first-half warrant loss was $74.7 million and Q2's was $205.6 million, so Q1 had a gain of about $130.9 million. COMPASS therefore reported roughly $91 million of net income in Q1 and a $254 million loss in Q2, while operating spending was $42.9 million in Q1 and $52.4 million in Q2. The company itself warns that this adjustment "can result in significant variability in our reported net income or net loss." For this company, look at operating loss and cash, not net loss.

What the headline numbers hide

• Cash burn matches the operating loss, which is a good sign. Operating activities used $88.2 million of cash in the first half, close to the $95.3 million operating loss. The gap is mostly non-cash stock-based pay ($6.9 million). The loss is not hiding extra cash outflows. One item to watch: "other assets" used $9.0 million of cash in the half, versus $1.4 million a year ago. The filing doesn't explain it in detail.
• Excluding warrants, the loss still grew faster than spending. Without the warrant charge, Q2's loss was about $48.2 million, compared with about $35.9 million a year earlier (our calculation from the income statement). That's up 34%, more than the 22% rise in operating loss. Two smaller items explain the gap. First, a $0.9 million foreign-exchange loss replaced a $2.3 million gain. Second, the UK R&D tax credit fell from $4.3 million to $2.6 million. The UK refunds part of R&D spending in cash. COMPASS now books the credit at the lower standard rate because it isn't sure it qualifies for the higher "R&D-intensive" rate, and the 10-Q says this is "currently under discussion with HMRC." Interest income on the larger cash pile rose to $3.7 million from $1.9 million, which partly offset these.
• Shareholders are being diluted heavily. Shares outstanding went from 96.1 million at year-end to 135.7 million at June 30, a 41% increase. More than 23.8 million pre-funded warrants are still outstanding on top of that. The loss per share grew less than the total loss ($0.41 to $1.88, or 4.6x, against 6.6x) only because the loss is now spread over more shares. Selling shares continued after the quarter: by July 30, at-the-market sales (selling shares directly on the stock exchange) and PIPE warrant exercises had raised another $31.7 million.
• The balance sheet looks better than it did. Shareholders' equity went from a $52.8 million deficit at year-end to positive $84.5 million. The company also borrowed $50 million from Hercules in January, under an amended loan that allows up to $150 million. Debt is now $50.7 million, all of it long-term.
• Guidance hasn't changed. The cash-runway guidance has said "into 2028" in the Q4 2025, Q1 2026 and Q2 2026 releases. Cash fell from $466.0 million at March 31 to $433.3 million at June 30, a $32.7 million drop that is net of a small amount of share sales during the quarter.

Takeaway: Ignore the $253.8 million loss. It is mostly a non-cash charge that grows when the share price rises. What matters is that COMPASS is spending about $45 million a quarter in cash and moving that spending from trials to launch preparation (G&A up 84%), with $433 million in the bank. Its outcome now depends on approvals from the FDA and the DEA, not on money. The cost is dilution: there are 41% more shares than in December, so each existing share owns about 29% less of the company.

The drug behind the numbers

COMP360 has positive results from two Phase 3 trials, the last stage of testing before approval. In the first trial, COMP005 (258 patients), a single 25 mg dose beat placebo by 3.6 points at six weeks on the MADRS depression scale, a 60-point clinician-rated score. In COMP006 (581 patients), two 25 mg doses three weeks apart beat a 1 mg comparison dose by 3.8 points at six weeks. In July 2026, the company reported 26-week COMP006 results. 39% of patients on 25 mg had a "clinically meaningful" improvement (at least a 25% drop in MADRS score) by week 6 and kept that benefit on average through week 26, compared with 25% after a single dose in COMP005. Serious adverse events were similar across arms (5.7% on 25 mg vs 6.3% on 1 mg).

These benefits are modest in absolute terms. A 25% drop in symptoms is a low bar for "response", and most patients didn't reach it. But this is a hard-to-treat group: participants had been in their current depressive episode for more than three years on average. The company's commercial argument is a few treatments a year instead of a daily pill. Whether doctors and insurers accept a multi-hour supervised session as practical will decide sales more than the trial numbers will.

What to watch next

• Regulatory filing: COMPASS is sending its New Drug Application (NDA), the formal request for FDA approval, in pieces under a rolling review, and expects the final part in Q4 2026. The FDA granted the drug a National Priority Voucher in April 2026. The company says this "has the potential" to shorten the review to 1-2 months. The program is new, and the 10-Q lists the risk that it "may not actually lead to a faster FDA review."
• DEA rescheduling: Psilocybin is a Schedule I controlled substance. After FDA approval, the DEA must reschedule it before the drug can be sold in the US, and states must follow. An April 2026 executive order tells the DEA to move quickly. The company says about 90% of US patients live in states that intend to reschedule within 30 days of the federal decision. Management targets a launch in the first half of 2027.
• Spending path: The 10-Q says expenses will "increase substantially" with launch, manufacturing scale-up and the PTSD trial (COMP202, a Phase 2b/3 trial cleared by the FDA in January 2026). We think quarterly operating losses will move above the current ~$52 million as the sales force is hired. That is probably why the runway guidance stays at "into 2028" even though cash has more than doubled since December.
• Possible extra cash: If the remaining 2023 PIPE warrants (exercise price $9.93) are all exercised, they would bring in up to $105.8 million. The company does not count them in its runway estimate.
• Next report: The Q3 2026 10-Q should arrive in early November, based on the May 13 and August 5 filing dates this year. Watch whether G&A keeps climbing at the Q1-to-Q2 rate and whether the NDA is still on track for Q4.