Keros Therapeutics (KROS) Q3 2026 earnings review
Takeda's $40 million payment makes Keros profitable, but spending climbs
A licence payment from Takeda made Keros profitable for a quarter, while the drug developer's own costs turned upward. The biotech recognised $40 million under that agreement, which turned a year-ago loss into a profit. Chief executive Jasbir Seehra pointed to “disciplined execution of our strategic priorities”, yet research spending rose 5%, its first increase since Takeda took over the development of elritercept. Keros also pushed back its estimate of how long the cash lasts, now into the second half of 2028.
| Total revenue | $40.0M $14.3M a year ago |
|---|---|
| Net income | $14.2M a $7.3M loss a year ago |
| Cash on hand | $259.5M $257.6M three months earlier |
| How long the cash lasts, the company's own estimate | Into the second half of 2028 previously the first half of 2028 |
⚖️ Verdict: 🟢 Bullish
The case got better, because the Takeda agreement paid out again and bought Keros more room. Cash finished the quarter above where it started, and the funding date moved two quarters further out, past the first readings from the muscle trial. One caution: the savings from handing elritercept to Takeda have run their course, research costs are climbing as the trial enrols, and half the licence payment was still unpaid on September 30.
The question now is what Keros will spend as the Duchenne trial fills up. One answer: the group runs the trial on roughly its existing cost base, and the funding estimate proves cautious. The other: enrolment pushes research spending well past that base, and the money runs short of the date the company has just named. Research spending in the year-end release is the first real reading.
🐂 Bull Case
Takeda Licence Revenue Put Keros in Profit
Keros has no approved medicine, so nearly all its money comes from Takeda, the drugmaker that licensed elritercept, the Keros treatment for low blood counts, in 2025. This quarter the biotech recognised $40 million of revenue under that agreement, against $10 million in the same quarter last year.
The payment reshaped the income statement. Keros earned net income of $14.2 million, where a year earlier it lost $7.3 million. Almost the whole swing traces to the licence line rather than to cheaper laboratory work. It is the first profit the group has posted after five quarters of losses.
- Profit per diluted share: $0.70, against a loss of $0.18 a year ago
- Revenue growth: 181% from a year ago
- Income from operations: $12.8 million, against a loss of $15.4 million
Why this matters: Takeda now runs and funds the late-stage work on elritercept, so money of this kind reaches Keros without new share sales. The other side is concentration, because one contract produces essentially all the revenue and its payment terms, not the laboratories, drive reported profit.
What to watch: the revenue line in the year-end release. Takeda pays in steps, and the June quarter brought nothing at all, so a second payment inside twelve months would show the partnership feeding Keros faster than its trials consume cash.
Cash Rose and the Funding Date Moved Out
Cash at the end of September stood a little above its June level, the first rise of 2026 after falls of $5.9 million and $23.9 million in the two quarters before it. Keros also moved its own funding estimate out, saying the money now lasts “into the second half of 2028”.
Takeda money did that, not thrift. Accounts receivable rose to $20.0 million from nothing at the end of June, so by our math half the licence revenue reached the bank and half remains owed. Without the collected half, cash would have fallen about $18 million.
What to watch: whether the rest of the money lands in the December quarter. A payment that slips would take cash use back towards $18 million a quarter, which turns a collection delay into a funding question three months from now.
First Patient Dosed in the Muscle Trial
Rinvatercept is the lead candidate at Keros, one of the proteins the group designs to correct faulty signalling in the TGF-β family, a set of proteins that govern how muscle, bone, fat, heart tissue and blood grow and repair. In September the first patient in its Phase 2 trial received a dose.
That trial treats Duchenne muscular dystrophy, an inherited disease that weakens boys' muscles from early childhood. Before starting it, Keros reported that Phase 1 testing in healthy volunteers produced “increases in muscle mass and bone mineral density” alongside less body fat. The company expects the first readings in the first half of 2027.
What to watch: those first readings, and before them any count of patients enrolled. Enrolment that keeps pace would hold the readout inside the window Keros has now named twice, which matters because this trial is the only clinical work the group still pays for itself.
🐻 Bear Case
The Takeda Cost Savings Have Run Out
For four quarters Keros looked like a company shrinking its way to safety. Handing the development of elritercept to Takeda and restructuring in 2025 cut total operating costs by roughly half against each year-earlier quarter. That comparison has now caught up with itself.
Costs of $27.2 million came in only 8% below last year's. Research spending turned up, to $20.5 million, and Keros tied the increase to “the additional research and development efforts to support the advancement of Keros’ pipeline”. The Duchenne trial dosed its first patient only in September, so most of its bill lies ahead.
- Total operating costs: 8% below a year ago, after four quarters of roughly 50% declines
- Research and development: 5% above a year ago, the first increase since the handover
- General and administrative: $6.7 million, a third below last year on smaller pay and fee bills
Chief executive Jasbir Seehra described “disciplined execution of our strategic priorities”, and the overhead line supports him. Yet savings in the back office cannot offset a trial that is only beginning to enrol, because research already takes about 75 cents of every dollar Keros spends, against 66 cents a year ago.
What to watch: research spending in the year-end release, set against the $20.5 million just reported. A bigger number while the Duchenne trial fills up would show the cost base growing with the pipeline, which shortens how long the cash lasts.
Without Takeda, There Is No Revenue
Everything Keros sells on its own now rounds to nothing. Service and other revenue came to $9,000 in the September quarter, against $4.3 million a year earlier, when the group still ran elritercept work that Takeda has since absorbed. Across nine months that line brought in $376,000.
Strip the licence revenue out and the September quarter was a loss of about $26 million by our math. Nine months into 2026 the biotech sits $38.3 million in the red, where the same stretch of 2025 showed a profit, because the Takeda agreement landed in one piece early that year.
What to watch: the revenue line in the year-end release. A return to a few thousand dollars would confirm that one partner's payment calendar, rather than any product of its own, decides whether Keros reports a profit in a given quarter.
Two Planned Updates Are Still Missing
Keros set two timetables before this quarter and added nothing to either. Early this year it planned to talk with regulators during the second half of 2026 about a Phase 2 trial of rinvatercept in amyotrophic lateral sclerosis, a nerve disease known as ALS. The November release names ALS as a target but skips the meetings.
- Regulatory talks on an ALS trial: promised for the second half of 2026, with seven weeks of the year left and no word
- Phase 3 trial of elritercept in myelofibrosis: a Takeda plan last described in the spring, still without a start date
- Cash flow statement: absent from the release, so the split between spending and working capital stays hidden until the quarterly filing
What to watch: the quarterly filing and any January presentation. Silence that carries into 2027 would mean the second use of rinvatercept has slipped a year behind the timetable Keros set out early this year, leaving one trial to carry the whole clinical story.
👓 Other Themes
KER-206 Steps Out of the Laboratory
Keros nominated KER-206 for further investigation, calling it “a novel ligand trap designed to bind select TGF-β ligands to regulate bone remodeling”. In plain terms, it aims to block proteins that let bone break down faster than the body rebuilds it. The company believes it could “increase bone mass and improve skeletal integrity”, and showed laboratory data in October. No trial exists yet.
💲 Other KPIs
The return Keros earns on its cash keeps shrinking: 68% less than a year ago, and smaller again than the June quarter. Last year's payout to shareholders took the cash balance down by more than half, so the same funds now earn far less each quarter.
Half of last year's 40.6 million, after the group repurchased and tendered 21.1 million shares for $384.6 million during 2025. Another 146,342 shares moved into treasury this quarter at a cost of $1.7 million. Because the quarter earned a profit, the diluted count now sits above the basic 19.9 million.
Costs paid in advance, mostly for research work, fell $16.3 million from the December level. Spending those prepayments put cash back on the balance sheet through 2026, a one-way help that is nearly used up: the balance is the smallest of the recent quarters.
🔮 Guidance
Raised. Keros now says its cash funds operations “into the second half of 2028”, where it had said the first half for four quarters. By our math the $259.5 million on hand must stretch seven more quarters. That allows roughly $37 million a quarter. The September quarter used far less, so the estimate looks cautious unless trial spending steps up.
Unchanged. Keros still plans to show the first readings from the muscle trial in the first half of 2027, the same window it named in August. The first patient took a dose in September, so dosing began about two quarters before that window opens. Nothing in this release moves the date either way.
❓ Key Questions
What triggered the $40 million Takeda payment?
The release ties the revenue to the licence agreement without naming the event behind it. Readers cannot judge whether similar payments follow, or when, until Keros describes the trigger.
When does the remaining $20 million arrive?
Half the licence revenue sat in receivables on September 30. The due date decides whether the December quarter shows another cash increase or a return to roughly $18 million of quarterly cash use.
Has Keros met regulators about the ALS trial?
The company planned those talks for the second half of 2026 and has said nothing with seven weeks of the year left. A Phase 2 design and a start date would show how fast the second rinvatercept programme moves.
How much will the Duchenne trial cost each quarter?
Research spending rose for the first time since Takeda took over elritercept, with enrolment barely started. A quarterly cost range for the trial would let readers test the funding estimate directly.
When does Takeda start the myelofibrosis Phase 3 trial?
Keros described that plan in the spring and has not updated it. The start date shapes future milestone payments, which are the only revenue the group currently receives.