Parabilis Medicines

IThe business

Parabilis makes medicines it calls Helicons: short protein fragments, stabilized into a helix so they keep their shape, that can get into a cell and stick to a flat surface on a protein inside it. Most drugs can’t do both. Pills get into cells but need a pocket to grip. Antibodies grip flat surfaces but stay outside. The company says about 80% of the proteins that drive disease are out of reach for that reason.⁴ It has one medicine in people, zolucatetide, and four programs in the lab. It has no approved product and has never had product revenue. In the quarter to June it booked $148,000 of collaboration revenue and spent $51.0M R (reported).²

The company started in July 2015 as FOG Pharmaceuticals, built on a license from the Harvard laboratory of Greg Verdine, and changed its name in October 2024.⁴ It has 145 employees, 122 of them in research and development and 69 with a medical or doctoral degree, plus 31 consultants.⁴ It lists on Nasdaq under PBLS. The listing priced on 9 June 2026 at $20 a share, trading began on 10 June, and the first close was $31.60.⁴¹ There is no ten-year record here. The company went public in June 2026, and as an emerging growth company it gave two audited years, 2024 and 2025. I use those, the first half of 2026, and the five months of trading that exist. Where I need more, I say what I am guessing.

The medicine and its target

Inside a cell, a protein called beta-catenin moves to the nucleus and binds a family of switches called TCF. That pairing turns on genes for growth. It is the last step of the Wnt pathway, which is stuck in the “on” position in more than 10% of cancers, either because beta-catenin itself is mutated or because a brake called APC is missing.⁴ Desmoid tumours are the cleanest case. They are rare, non-spreading but invasive growths of connective tissue, driven by beta-catenin or APC changes, and they peak between the ages of 30 and 40, more often in women. The company counts about 11,000 people in the U.S. who are actively managed by a doctor, out of about 30,000 living with the disease.⁴⁵ Drug makers have tried for about 30 years to block the beta-catenin and TCF pairing and not got there; the company says zolucatetide is the first drug to do it directly.⁴

Zolucatetide is given by infusion: weekly for four cycles, then every second week.⁵ The company plans an injection pen, but it will run the Phase 3 trial on the infusion, and says it may need extra studies to show the pen works the same way.⁴ Because the drug blocks a pathway the body uses, it has side effects of its own. Bone loss is a known risk of Wnt blockers, so patients take a bone-protecting drug throughout. Blocking the pathway also lowers the hormone aldosterone, which can cause fatigue and low sodium; some patients take a steroid pill for it. The drug also raises bilirubin without harming the liver, which the company compares to a harmless inherited condition.⁴

What the early data show

The data come in two cuts. The first, to February 2026, is in the prospectus: 38 patients treated, 25 with at least one scan. All 25 had smaller tumours. Of 19 with two scans, 14 had a response, which is a shrinkage of 30% or more, and one tumour disappeared. The second, with a data cutoff of 22 June 2026, was presented on 24 October at a cancer congress in Madrid and reported in a company release: 76 patients in all, 48 of whom started at 450 mg or more, the dose the company now favours, and 46 of those could be scored. All 46 had tumour control, and 13 of the 18 who had reached the 32-week scan had a response, or 72.2%.⁴⁵ The presenting doctor, Mrinal Gounder of Memorial Sloan Kettering, is quoted in that release as calling the results “encouraging, particularly the magnitude and apparent rapidity,” and the release says he has financial interests related to the company.⁵ I read the release on a syndicated copy, not on the company’s own site, and I did not see the slides.

I read these numbers as promising and small. Three things limit them. First, the numbers of patients are tiny. The second cut counts 18 patients at the 32-week mark, from a group treated for a median of only 4.7 months, and the release itself says the cohort is recent.⁵ The February count was 19, so the share of patients responding has not moved while the denominator has changed. Second, the release gives no length of response and no progression-free survival. Those are what the Phase 3 trial will measure; the company says the FDA has aligned with it on a placebo-controlled trial with progression-free survival as the primary endpoint.⁵ The prospectus said it would also ask for response rate as a second primary endpoint. The 24 October release does not mention that, so I treat it as open.⁴⁵ Third, the two pills were measured over a long time. Ogsiveo’s 41% is the rate at the end of a trial in which responses kept arriving for a median of 20.6 months, so a young trial with a high rate and a finished trial with a lower rate tell me less than the gap suggests.⁴ A fair summary is that the first look shows a higher rate than either pill’s, and that it is a first look.

The market, and who is already in it

One drug is approved for desmoid tumours. Ogsiveo is a pill that blocks an enzyme called gamma-secretase. Merck KGaA owns it through its purchase of SpringWorks, which closed on 1 July 2025 at an enterprise value of $3.4B.¹¹¹² In the six months to December 2025 Merck KGaA booked €134M of Ogsiveo sales, about €270M a year if that pace held, which is my arithmetic C (calculated).¹¹ A second pill from the same family, varegacestat from Immunome, was filed in April; the FDA’s decision date is 28 April 2027.⁹ Both cause diarrhea in more than 80% of patients and ovarian damage in half or more of women able to have children, and about one patient in five stops because of side effects.⁴ The company says many patients delay or avoid treatment because of those side effects, and others take older chemotherapy or anti-inflammatory drugs. Its argument is that a gentler drug would lower the bar for starting treatment and make the market larger than it looks.⁴ That is a reasonable argument and I can’t test it.

What the pills earn tells me the size of the prize. SpringWorks launched Ogsiveo in November 2023 and booked $172.0M of U.S. net sales in 2024 and $44.1M in the first quarter of 2025.¹³¹⁴ After Merck KGaA took over, the drug sold €134M in the second half of 2025, about €270M a year, or roughly $310M if a euro is worth $1.15 E (estimate); that figure includes sales outside the U.S.¹¹ I found no net price per patient in any filing I opened. If it is $200,000 to $300,000 a year, an assumption of mine E (estimate), those sales mean about 1,000 to 1,600 patients on the drug, or 9% to 14% of the 11,000 the company counts E (estimate). SpringWorks put new U.S. cases at 1,000 to 1,650 a year, so the 11,000 is mostly a stock of existing patients that a new drug draws down, not a yearly flow.¹³ My base case has zolucatetide reaching 2,400 U.S. patients, 22% of the 11,000 and about one and a half to two and a half times Ogsiveo’s count at the end of 2025. That is more than the incumbent has reached in about two years, and it rests on the idea that a gentler drug widens the market.

The rest of the pipeline

Zolucatetide is also in a study of familial adenomatous polyposis, an inherited condition in which the large bowel fills with polyps, about 34,000 people in the U.S. by the company’s count, with no approved medicine.⁴ Two patients with both polyposis and a desmoid tumour saw their polyps shrink; in one, the duodenum went from the most severe grade, Spigelman stage IV, to near normal.⁴¹⁵ The company announced on 14 October that a dose-ranging polyposis study had started, and more data are due in the first quarter of 2027.¹⁵⁸ In liver cancer, one heavily treated patient with a beta-catenin change had a partial response for nine months, and the company opened a cohort on the strength of it.⁴ In colon cancer the company reports stable disease and falling tumour DNA in some patients, and is testing combinations.⁴ Those are anecdotes. Beyond zolucatetide there are two prostate-cancer degraders, which break down a protein and not only block it, and a degrader for beta-catenin. None has been in a person.⁴

The Regeneron deal

In May 2026 Regeneron signed a research deal for antibody-and-Helicon conjugates on five targets, mostly outside cancer. Parabilis received $50M up front in June and Regeneron bought $75M of stock in the listing at $18 a share, 10% under the IPO price. The company could earn up to about $2.2B in milestones ($470M for development, $575M for approvals and $1.15B for sales) plus royalties from the high single digits to the low double digits. Regeneron controls everything after the lab stage.⁴² The headline number is not what I expect to receive. I expect most of that headline never to be paid, and I use 4% of it in my model E (estimate). The deal still tells me a large company paid to look at the platform. It also limits the company: it may not work on competing products against those five targets.⁴

What happens next

These are the dated events I can see.^4816917 The third-quarter report has not been filed. The deadline is Monday 16 November, 45 days after quarter end with the weekend rolled forward, so I mark mid-November as expected, not confirmed.¹⁸³ The company said on 24 October that it had reached alignment with the FDA, after end-of-Phase 2 talks, on the dose, a placebo-controlled design and progression-free survival as the primary endpoint.⁵ The lock-up on nearly all pre-listing holders, and on Regeneron’s private-placement shares, ends 180 days after the 9 June prospectus. That is Sunday 6 December, so the first day those shares can trade is Monday 7 December.⁴ More polyposis data and data on the rare brain tumour are guided to the first quarter and the first half of 2027.⁸ The registry shows primary completion of the main Phase 1/2 study in April 2027 by its own estimate. Immunome’s decision date is 28 April 2027. The company guides to a Phase 3 start in the first half of 2027 and topline results in 2029.

The company’s cash plan in the quarter to June read “into 2030.” Its prospectus, written before the underwriters bought extra shares, said the second half of 2029.²⁴ My own path lands between the two, and I show the range in Financials.

IIThe moat

A protected business is one where a rival can’t copy the product and take the customers. Parabilis has three possible sources of protection: patents, being first, and a method for making these molecules. I looked for each in the filings. None is proven yet.

Patents

As of March 2026 the company owned or licensed at least 17 patent families, 12 of them its own. The families that cover zolucatetide itself run to at least 2042, and families on how it is made up or used run from 2042 to 2044. No patent had been issued from any of those families. They are applications, in the U.S. and in other countries.⁴ An application can be narrowed or refused. If they issue, the law lets a company add up to five years to one patent, but not past 14 years after approval; if the drug is approved around the start of 2031 that cap is 2045. In my base case protection ends in 2044, in my bear case in 2042 with no extension, and in my bull case in 2045 E (estimate).⁴ Orphan status gives seven years of exclusivity after approval, but only against the same molecule. It did nothing to stop varegacestat, which is a different one.⁴ Twelve of the 17 families are the company’s own and the rest are licensed. The Harvard license costs low single-digit royalties, up to $18.3M in milestones per product, and a low-to-mid-teens share of partner income of certain kinds; the prospectus does not say whether Regeneron’s $50M falls under it.⁴

Being first

The company says zolucatetide is the first drug to block the beta-catenin and TCF pairing directly, and the prospectus names no other drug in that class.⁴ I could not check that against another source, and a first-in-class drug can be passed by a better second drug. Being first is useful only if the first drug works. In desmoid tumours it has two pills to beat, and the company is clear that the comparison will be made on response, speed, side effects and how the drug is given.⁴

Making the molecules

The company says a decade of work produced millions of data points on hundreds of thousands of these peptides, and that this feeds models that design new ones.⁴ It uses contract manufacturers and owns no plant, and the prospectus warns that it relies on single suppliers for some materials.⁴ That is a claim I can’t test. The test is the pipeline. One drug has reached people. The next two programs have not, so the method has one result so far.

My reading: no protection that shows in the numbers yet. The biology is hard, 30 years of effort by others did not get a drug into patients, and the company has a clinical result nobody else has. That gives it a head start. A rival with a good pill could catch up, and no patent has issued to stop one.

IIIThe competitive advantage

The first section asked who is stopped from competing. This one asks what Parabilis does better than the people it competes with. The answer is one thing, shown in a small trial: it appears to shrink tumours at a high rate, quickly, without the damage the pills do to the ovaries.

That last point matters more here than it would elsewhere. Desmoid tumours peak between 30 and 40 and mostly in women; 69.7% of the 76 patients in the June cut were female, and the median age was 37.⁵ A pill that damages the ovaries in half or more of women of that age gives a doctor a reason to wait, and the company’s case is that the reason to wait is a reason to treat later and in fewer patients.⁴ If a gentler drug moves treatment earlier, the market grows. That is the main route, in desmoid tumours, by which zolucatetide could be worth more than the pills’ sales suggest, and the evidence for it is one release and one prospectus.

What works against it

Three things work against the advantage. The first is how the drug is taken. Weekly infusions for four 28-day cycles, then every second week, are a heavy ask of a young adult who can take a pill at home. The company knows this and is developing a pen, but the pivotal trial uses the infusion, and it expects the FDA to want studies showing the pen gives the same drug levels, plus tests of whether patients can use it unaided. Any of those could delay approval of the form it wants to sell.⁴ The second is the side effects that come from the mechanism. Blocking the pathway can thin bones and lower aldosterone, so trial patients take bone drugs and some take a steroid.⁴ A doctor will weigh those against diarrhea. The third is timing. Varegacestat has a decision date in April 2027, nearly four years before I expect zolucatetide to be approved, and Ogsiveo will have had seven years on the market by 2031 E (estimate).⁹

Awards and designations

The FDA has given the drug Fast Track status and Orphan Drug designation for desmoid tumours.⁴ Fast Track lets the company send an application in pieces and talk to the agency more often, and it can lead to a priority review. It does not change the standard for approval. I count it as a small help for timing and nothing about the odds.

The platform case

The company’s larger claim is that zolucatetide proves a method. Regeneron’s $125M of payments and stock purchase is a vote of some confidence in it, though the deal is for outside oncology, on five targets, and pays most of its headline only if things go far.⁴² I give the platform a modest value. A platform that has produced one drug in people, after 11 years, is not the same as a factory, and I list it as the thing I can’t judge best. If the ERG or ARON degraders reach people and show activity, I will revise this section.

IVManagement

Who they are

Mathai Mammen has been chief executive and chairman since June 2023. Before that he ran research and development at Johnson & Johnson from 2017 to 2022, was a senior vice president at Merck and co-founded Theravance, where he led the research teams. He has sat on the Sandoz board since May 2024. The company credits him with leading the discovery or development of 19 approved medicines.⁴ That is a record of picking drugs and running large programs. At Parabilis he has not yet taken a drug through a Phase 3 trial. The chief financial officer, Thomas Kotarakos, has been at the company since 2021 and took the job in February 2026. Helen Ho, who ran business development at Blueprint Medicines, joined in April.⁴ In August the board added a ninth director, Craig Tendler, with an option struck at $39.36, the Friday close.¹⁹¹

The weak spot is medicine. Keith Orford, the chief medical officer, resigned in April 2025 with a $601,000 severance. Fawzi Benzaghou replaced him in September 2025 with a $115,000 sign-on bonus. On 28 August 2026 the company said he and the company “mutually agreed” he would leave on 30 September, with severance.⁴¹⁰ Two chief medical officers have gone in 18 months, and the one in the job when the June data were cut left before they were shown. In October the company said it had reached alignment with the FDA on the Phase 3 design.⁵ The filings I read name no replacement for the medical chief.

How they are paid, and what that pays them to do

Dr Mammen’s 2025 pay was $1.30M: a $784,840 salary and a $496,411 bonus. The bonus is discretionary and tied to clinical, financial, organizational and pipeline goals, which the board scored at 115% of target.⁴ He received no new options that year. His real stake is the options he already has. In April 2023 he got 2.8M options struck at $3.43, with 419,506 of them earned only on performance targets, and in March 2024 he got two more grants struck at $1.48.⁴ At $30.37 the 2.79M he can exercise are worth about $85M before paying the strike, my arithmetic C (calculated). The pay points him at the share price and at keeping the company funded, which means the next trial result and the next financing.

The wider pattern is a large option pool granted at low strikes. In the first half of 2026 the company granted 7.4M options at an average strike of $3.82, a few months before it listed at $20. At $30.37 those grants carry about $197M of gain C (calculated). In all, 15.7M options were outstanding at 30 June, 12.7% of the shares, at an average strike of $3.23, and the plan has a further 11.1M shares reserved plus 1.1M for an employee share-purchase plan.²⁴ The accounting cost so far is small, $4.9M in six months, because the options were valued before the listing. It will rise. I treat the pool as a cost to owners and count the options, net of their strike, in the share count (137.6M, against 123.5M outstanding).

Their own money

Almost none of the managers’ money is in shares they bought. The 13 officers and directors as a group were counted as holding 3.49M shares after the listing, 2.87% of the company, and 3.37M of those are options.⁴ Director Alan Sebulsky bought 12,500 shares in the listing at $20 and 17,500 on the market at about $27 in June.⁷ I read every Form 4 on the company’s list through August. None shows a sale. They cannot trade before the lock-up ends in December, so the absence of sales says little.

What they did with capital

The company has not had a full cycle. In eleven years it has lost $639M and raised $877M before the listing and $788M in it.²⁴ It dosed more than 150 patients with the one drug and has data on 76 in desmoid tumours.⁴⁵ It raised its last private round, Series F of $305M, in January 2026 at about $9.48 a common-equivalent share C (calculated), then a $50M SAFE, a contract that converts into shares, which converted at $18, then the public offer at $20.⁴ Those steps took the price up 2.1 times in five months. They show a company that could raise on better terms each time, which is a skill, and which depends on the market staying open.

Who else owns it

RA Capital, a Boston investment manager, bought 21.2M of the 38.5M shares sold in the listing, 55% of them, for $423.8M at $20, and its filing in June put its stake at 28.9M shares, 23.8% of the voting shares; with its July purchases it holds 29.2M, 23.6% of all shares outstanding in August C (calculated).⁶ It added 245,000 more shares on 20 to 22 July at $27.13 to $30.00.⁷ Fidelity’s parent, FMR, reported 10.3M shares, 8.5%, at 30 June, up from 8.8M before the listing, and ARCH Venture Partners reported 6.4M, 5.3%.²⁰²¹⁴ The other large holders are from the prospectus and so from before the listing: venBio 5.2M, GV 4.8M, Milky Way Investments 4.6M, Cormorant 4.4M and Nextech 4.2M. Any of them may have bought more in the listing, which the filings I read do not show.⁴

Alphabet’s 13F filing for the quarter to 30 September 2026 lists 4,919,498 Parabilis shares, worth $158.85M at the quarter-end, with GV Management Company named as the other manager. The prospectus shows GV funds holding 4,833,248 shares before the listing, so the 13F is 86,250 shares higher, and a GV general partner, Krishna Yeshwant, has been a director since April 2018.²²⁴ A 13F shows what a firm held on one date and says nothing about how it sees the stock. Regeneron’s own 13F lists the 4,166,666 shares it bought in the private placement, worth $134.5M on 30 September.²³

Two other points about the people in the deal. The first is who ran it. Leerink Partners, BofA Securities, Evercore, Guggenheim and LifeSci Capital were the listing’s bookrunners.⁴ Jeffrey Leerink, the chairman and chief executive of Leerink Partners, was a Parabilis director until May 2026, bought $275,000 of Series E and $550,000 of Series F stock, and got an option on 71,479 shares in return for advisory services.⁴ The second is the float. The prospectus lets a locked-up holder sell shares it bought in the listing or on the market afterwards. On my reading, that frees RA Capital’s 21.2M listing shares and its 0.2M July purchases; its 7.7M pre-listing shares stay locked.⁴ RA’s 13D describes a 180-day lock-up on its funds, with limited exceptions, and does not list them, so I rely on the prospectus for the exception.⁶ That puts about 38.8M shares in free hands now, 31% of the total, and 55% of those are RA’s C (calculated). The other 80.5M held before the listing, plus Regeneron’s 4.2M, are locked until Monday 7 December: about 85M shares, or 69% of the total C (calculated).⁴

VFinancials

There is no ten-year history to show. The prospectus gives two audited years, 2024 and 2025, because the company is an emerging growth company, and this page uses those, the first half of 2026 from the 10-Q, and the five months of trading since the listing.⁴² The company has no sales and a single line of revenue, $148,000 from Regeneron in the second quarter. The question is what it costs to run, how long the money lasts, and what each round of money cost owners.

Spending has grown fast and will grow more. Research and development rose 25% in 2025 to $125.6M and 20% in the first half of 2026. Of the 2025 research spend, $41.8M went to zolucatetide directly, $7.4M to the ERG degrader, $4.0M to ARON and $1.1M to the beta-catenin degrader; the rest was people, labs and rent.⁴ General and administrative cost rose 68% in the first half; the company put the first-quarter rise down to listing preparation and patent legal fees.⁴ I expect the total to go from about $218M this year to near $270M in 2027, because the Phase 3 trial starts in the first half and the company is building the staff to run it E (estimate). On 30 October it extended its Cambridge lease to the end of 2037 and took 33,525 more square feet from 2028, with rent starting in 2029.²⁴

What an owner can take out

I value the cash an owner could take out after stock pay and after the money needed just to keep the lab and offices going. Here that cash is negative every year. In 2025 the net loss was $145.9M and operating cash flow was −$123.7M, because $13.1M of costs were non-cash and unpaid bills grew by $13.6M, partly offset by $6.3M of lease payments.⁴ After capital spending and the $3.3M of stock pay that I count as a cost, owner cash was −$127.3M C (calculated). In the first half of 2026 owner cash was −$100.2M after taking out $41.5M that came from Regeneron’s up-front payment and was booked as deferred revenue. That is a prepayment for work, not earnings. So the business burned about $100M in six months, and about $50M a quarter.

The company ended 2025 with $27.7M. It was spending about $10M a month, so it was three months from the end. That changed in the first quarter, when it raised $305M of Series F stock and a $50M SAFE, and then the listing brought in $788M after costs. At 30 June it had $1,120.7M in cash and securities.⁴² The third-quarter report is not out; EDGAR showed no 10-Q for the quarter to 30 September on 10 November, and it is due by 16 November.³¹⁸ So the 30 September cash is my estimate, and I give a range. Operating cash burn was $49.2M in the first quarter and about $54M in the second, once Regeneron’s $50M up-front payment is taken out C (calculated).⁴² I expect interest on $1.1B to add about $11M a quarter E (estimate). I use $1.06B for 30 September, which is the June balance less about $60M. That is a little above the recent run rate of burn, capital spending and loan repayment, so it sits at the cautious end of a $1.04B to $1.08B range E (estimate). Each $20M moves my base value by about 15 cents. The company says its cash is enough “into 2030.” My own path in the chart spends about $62M in the fourth quarter, $270M in 2027, $300M in 2028 and $320M in 2029. That leaves $197M at the end of 2029 before any new money, or $175M to $219M across the cash range E (estimate). At the pre-launch spending I model for 2030, about $71M a quarter, that is a little under three quarters of cash, so it runs out around the third quarter of 2030. That fits “into 2030” and is later than the prospectus’s “second half of 2029.” In my bear case, with spending 10% higher, $76M to $121M is left, one to one and a half quarters of cover.

What the rounds cost owners

Before the listing, owners of the preferred shares had paid an average of $10.74 for each common-equivalent share. New buyers in the listing paid $20, and the prospectus put the immediate dilution of net tangible book value at $11.64 a share.⁴ Today’s buyers pay $30.37.

The cost of the next round is the part I model. If the Phase 3 trial succeeds, the company needs roughly $300M to $500M more before launch revenue covers its costs, depending on how well the drug sells E (estimate). I assume it sells shares at 12% under fair value to raise it. The cost of that discount, 13.6% of the amount raised, comes out of my value. If the trial fails, it does not raise.

The share price

The stock opened at the listing, closed its first day at $31.60, dropped to $25.01 within three days, and then went up and down in a wide range. Realised volatility, from the daily closes, is about 76% a year C (calculated). The trading range for the five months is $22.75 to $41.73 on closing prices.¹ It fell from $41.73 on 18 September, a day on which 9.1M shares traded, several times the usual, to $22.75 on 16 October, down 45%. I looked for a company announcement that explains that fall and found none. The company’s news in those weeks was the date of its congress talk and the start of the polyposis study, and neither explains a drop of that size.¹⁶¹⁵ I do not know what drove it, so I draw no conclusion from it about what the market knows. After the ESMO data on Saturday 24 October the stock rose 2.4% on Monday, and the 9 November close of $30.37 is 24% above the Friday close before the data, $24.44.

VIValuation

At $30.37 the price is 461% above my base value of $5.41 and 13% above my bull value of $26.83. My margin-of-safety price is $1.04. The price needs the drug to work in common cancers, and I put the odds of that near 7% if the desmoid drug is approved. This section shows how I got there, where I am unsure, and what the price would need instead.

First, how this goes badly wrong

Before any number I asked what would most likely destroy this investment, and worked back from that. Three things stand out.

The Phase 3 trial fails, or wins too narrowly to sell. This is the main event. I give it 40% in my base case. The oncology record for drugs that reach a Phase 3 trial is worse: 47.7% went on to a filing and 92.0% of filings were approved, a 44% chance of approval. For rare diseases of every kind the record is better, 56.6%.²⁵ The company says it has an agreed design with a placebo, and two rival drugs have already beaten placebo in this disease. Against that, its own evidence is early, and the rivals block a different enzyme, so their wins do not carry over. I set approval at 60%, a little above the rare-disease record, and use 40% and 80% in my bear and bull E (estimate). A failure costs more than people think. By the end of 2029 the company will have used about $863M of its cash on the way to find out, net of interest, and the $1.06B that looks like a floor of $7.70 a share is $1.43 a share by then E (estimate).

The money runs short at the wrong moment. If the trial succeeds, launch costs, a sales force and continued research push the company into needing $300M to $500M more. A success would let it raise, but I assume it sells at a 12% discount, and if markets are shut it sells cheaper or partners the drug. A failure does not need money, but a long delay does. The company’s own cash plan says “into 2030,” and the Phase 3 result arrives in 2029 E (estimate).²⁵

It works, and it sells less than the story needs. Desmoid tumours are rare. Two pills will be on the market. A weekly infusion is harder to take than a pill, and a doctor may keep the new drug for the patients the pills have failed. At 1,200 patients, my small launch, the drug earns less than it costs to market and the stock is worth about what the failure case gives.

None of these needs anything exotic. They set my bear case, and they are why I do not start from the cash balance.

What an owner could really take out

No cash can be taken out of this company for at least a decade, if ever, so the usual question changes. It becomes: how much does the company have to spend before the facts are known, and what is the chance the facts are good? Owner cash in 2025 was −$127.3M after stock pay and capital spending, and about −$100M in the first half of 2026, as the Financials section shows. That is far from reported earnings, because there are none. I model the spending year by year: $62M in the fourth quarter, then $270M, $300M and $320M in 2027 to 2029. Of the $952M, about $508M goes to zolucatetide, $290M to the degraders and discovery, and $154M to running the company E (estimate).

What the price implies

The market values the whole company at $4.18B on my share count. Take away the cash and what is left is $3.12B, which is what the market pays for the pipeline. My value for the whole company is $5.41 a share, which is less than the $7.70 of cash a share, because the cash is mostly used before the answer is known. The market pays $30.37. The gap is $24.96 a share, or $3.4B across 137.6M shares, and it is a payment for outcomes I can’t see in the data.

The two ways to reach the price are plain. One is near-certainty on desmoid tumours: with every other assumption at my bull case, the approval odds would have to be 92%. The other is a successful program in the common cancers. With my bull sizes, the price needs 31% odds of an $8B drug in colon or liver cancer. The drug has treated a handful of patients in each. I think odds that high are more than the evidence supports, and I say that knowing larger investors put hundreds of millions into the listing at $20. They may have information I lack.

What I can’t judge

• Whether the Phase 3 trial succeeds. The data are 76 patients, no controlled comparison, and a median of 4.7 months on treatment. Anyone who gives you a precise number here is guessing too.
• How many patients take it. At the price I assume, Ogsiveo’s sales point to roughly one patient in ten of the 11,000 the company counts, about two years after launch. I assume zolucatetide reaches 22%, with a range from 11% to 33%. A pen would help; I can’t tell how much.
• Colon and liver cancer. Stable disease in some colon patients and one liver response are not a base rate. My 7% is a guess with a wide band, which the sensitivity tables show.
• The platform. A decade of design data, a partner paying $50M, and no second drug in people. I value it low and show it separately.
• What protects the drug. No patent has issued. I assume 2044 and show what 2042 and 2046 do.
• The price of the drug and the cost of making it. The company gives neither. I found no net price in SpringWorks’ or Merck KGaA’s reports either. I take $290,000 a patient-year net, my own estimate, and the cases show $250,000 and $330,000.

Because I can’t know these, I show a range and the price’s place in it. A fair reading of my work is that the base value is somewhere between $3 and $9, and that nothing in it gets near $30 without the common cancers.

How the model works

I value the company as the odds-weighted sum of what happens. Each year I count cash in and out: sales, less cost of goods (15%, including a royalty to Harvard), less selling and administrative cost, less research, less tax at 25% after using the company’s losses (about $560M by the end of 2026, plus the spending to 2029 E (estimate)). Where cash runs out, I add the cost of raising it at 12% below fair value. If the trial fails, I take a quarter off the cash that is left, for the cost of winding down. I run it to 2050 and let sales fall as protection ends. I take the spending to 2029 as certain, because the trial must run to find out. Cash is added at its face amount and the spending is taken away at the same 12%. If I discounted the spending at the Treasury yield of 4.88% instead, my base value would be $4.38, because the spending would weigh more.²⁶

The three cases

Bear

The trial has a 40% chance of success, the launch is a year late (2032) and small, the drug sells at $250,000 a patient, spending is 10% higher, protection ends in 2042, and the other programs cost more than they return. Most outcomes end in about what is left of the cash. Value $0.78.

Base

The trial has a 60% chance. Sales start in 2031 and reach $835M in 2038 for the middle launch. Polyposis works one time in four, a large solid-tumour franchise one time in 14 (7%). Protection runs to 2044. Value $5.41.

Bull

The trial has an 80% chance, approval comes in 2030, the drug sells at $330,000 and reaches 3,600 patients in the middle launch, polyposis works half the time, and an $8B solid-tumour franchise has 20% odds. Protection runs to 2045. Value $26.83.

Look at what the chart shows. If the drug is approved and sells at the middle launch size, a share is worth $7.34 on my model, a quarter of today’s price. Even the broad launch, with sales above $1.2B a year, gives $12.69. This is the arithmetic of a drug for a rare disease. The company has to run a Phase 3 trial and build a sales force, and its cash flows come late and are discounted at 12%. Then it sells a few hundred million dollars of a product that loses protection in 2044.

Where the value comes from

Cash of $7.70 a share is the biggest piece, and the spending to 2029 takes 5.25 of it back. The rest of the value is the odds of success. The desmoid drug itself is worth about $2.86 a share net of its own launch cost and the failure case. Polyposis and the solid tumours add $0.74 between them, weighted by odds, and the degraders and Regeneron deal come to −0.64 after what they cost. I charge the degraders their full spending, about $85M to $95M a year to 2029. On the odds I use, the three degrader programs together are worth less than they cost. A reader who thinks a degrader is likelier than 4% to work, or likelier to be partnered early, will get more.

What moves the value

Two things stand out. Approval odds move the value less than I expected: raising them from 60% to 100% adds $2.56 a share, because success is spread over a long road of costs and a patent clock. The solid-tumour inputs move it more: 25% odds on a $10B drug is worth $12.31. One at a time, each of these changes the base: launch a year later $4.58 and a year earlier $6.59; U.S. patients 25% lower $3.98 and 50% higher $8.54; net price $250,000 $4.63 and $330,000 $6.19; protection ending in 2042 $4.84 and 2046 $5.86; no value for the degraders and Regeneron deal and no spending on them after 2029, $6.05. None gets near the price.

The discount rate

I use 12%. That is 4.88%, the 10-year Treasury yield, plus about seven points for the risk of owning a company with one drug in the clinic.²⁶ I could not measure a beta from five months of trading, so I use the rate I used for Iovance, which has similar risks.²⁷ At 10% my base is $6.45, at 8% $7.89 and at 14% $4.67. I apply 12% to every year, including the years after approval, when most of the drug’s risk is gone, and to cash flows I have already weighted by odds. That is cautious. A buyer such as a large drug company discounts at a lower rate than an owner of a single stock, and the cross-check below tries that.

Where my numbers may be too low, and where too high

Too low. Four of my choices lean cautious. The first is the discount rate: at 8% my base is $7.89, and at 10% $6.45. The second is the platform. I count three named degraders against the whole research budget after 2029, and a method that keeps producing programs would be worth more than three. With no value for it and no spending on it, the base is $6.05. The third is the common cancers: at 15% odds on a $6B drug the base is $7.44, and at 25% odds on a $10B drug it is $12.31. The fourth is the buyer’s view, below. None of these reaches the price alone.

Too high. My approval odds of 60% sit above the oncology record; at 44% the base is $4.38. I also charge nothing for stock the company will grant from now on. The plans reserve 12.2M shares, and if they were all granted they would take up to about $0.44 off the base. Going the other way, I count the 15.7M existing options as if exercised at today’s price, which is cautious: at my base value many of them would be out of the money, and counting them at that value lifts the base to $5.73. I keep the cautious count, and the two roughly cancel. Costs after 2029 are my guesses, and a pen that needs extra studies would add to them.

Two cross-checks

The first is a purchase price. In 2025 Merck KGaA bought SpringWorks, which had Ogsiveo and a second approved drug, for an enterprise value of $3.4B.¹² The two drugs sold €189M in the six months after Merck KGaA took them over, about €380M a year C (calculated).¹¹ That is about 8 times annual sales, for products already on the market, with a pipeline thrown in. If I say a buyer would pay 8 times the third-year sales of each approved case, the base value rises by $2.59 to $8.00 E (estimate). My model, valued at approval at 12%, gives $1.33B for the middle launch and $1.77B at 8%, compared with $2.24B at 8 times sales E (estimate).

The second is the reverse. If the price is right, what has to be true? On my model, with my bull case for everything else, it takes either 92% approval odds or 31% odds in the common cancers. I think that is a high bar for a drug with no controlled data. I have not opened any sell-side research on the stock, so I use none of it here. Any analysts who cover it may have seen more than I have, and a price above my bull value means the market is paying for something the data I can read do not show.

Other uses of the same dollar

The nearest comparison on this site is Iovance, a company with an approved cell therapy that is trying for lung cancer. At its price I have the stock 40% above my base value.²⁷ AbCellera, with one drug that has a positive Phase 2 result, is 271% above.²⁸ Parabilis, at 461% above, has the biggest gap of the three. The next best use of a dollar I can put a number on is a 10-year Treasury at 4.88%, which does not carry the risk of a total loss.²⁶ I am not saying any of these is a good use of a dollar. I am saying that on my numbers this price asks for more than the other two ask.

The margin of safety

I set it from the bear case: I want a price at which my bear value is at least three quarters of what I pay. That price is $1.04. The price is 29 times that. A margin of safety needs value that does not depend on a good outcome, and for a drug company the nearest thing is cash. Here the cash is committed to a three-year program, and in my bear case the company has $98M left at the end of 2029, about 71 cents a share. So the margin of safety is a measure of how little is left if things go wrong, not a price I expect to see.

Two different questions

The first is how good the business is. It has a first-in-class drug with a promising first result, a chief executive with a long record of approved drugs, $1.06B in the bank E (estimate) and a large partner in Regeneron.⁴⁵ It also has no patents issued, two medical chiefs gone in 18 months, and one drug. The second is how good the price is. At $30.37 the market value is $4.18B for a company whose cash is committed and whose first product would sell, in my base case, $835M a year. Those are different answers. A promising business can still be a poor purchase when the price already pays for a big success. This looks like that case, and the strongest argument that I am wrong is in the next section.

VIIThe case for the price

What most people miss

Three facts cut in the company’s favour. First, speed: the median time to a response was 3.8 months in the February cut, against 5.6 on Ogsiveo and 8.2 on varegacestat.⁴ Second, the patients: 69.7% of the 76 treated were women, median age 37, and the pills damage the ovaries in half or more of women able to have children, so a drug that spares them has a reason to be used first.⁵⁴ Third, who bought. RA Capital paid $73.5M for preferred stock before the listing, then $423.8M for 21.2M shares at $20, almost six times as much, and bought 245,000 more in July at $27 to $30 C (calculated).⁴⁶⁷

The bull case

No named fund manager has published a case on Parabilis that I could check against a dated document. EDGAR shows 35 13F filings naming it from 1 October to 9 November, none from Altimeter, Atreides, ARK, Pershing Square or Coatue, and many are not due until 16 November.²⁹ The company calls zolucatetide “a pipeline in a product,” because one pathway drives many tumours.⁸ Dr Mrinal Gounder is quoted in the company’s release calling the results “encouraging, particularly the magnitude and apparent rapidity”; the release notes his financial interests.⁵ Holders have spoken with money. Fidelity’s parent, FMR, reported 10.3M shares, 8.5%, at 30 June, up from 8.8M before the listing, and Alphabet’s 13F shows GV with 4.9M shares at 30 September, 86,250 more than before.²⁰⁴²² None has said why in a document I opened, and a position is not a view. I give no analyst targets because I opened no sell-side research.

The case in plain words: a drug that shrinks tumours fast, without ovarian damage, makes a small market larger. The market may be bigger than the 11,000 figure, since about 30,000 people live with desmoid tumours in the U.S.⁵ The same drug then works in other tumours driven by the pathway: polyposis first, about 34,000 people, then in time liver and colon cancer.⁴ The company also counts 5,000 to 9,000 U.S. patients with a rare brain tumour, ACP, with data due in the first half of 2027.⁸ Holders of that view would say today’s price pays for only part of it.

Against reality

If the trial succeeds at the broad size and the common-cancer program is real, the price is possible. That is my bull case, $26.83, and the price is 13% above it. If everything works at my bull sizes (a broad desmoid launch, polyposis approved, an $8B drug in colon or liver cancer), a share is worth about $78 E (estimate). Against failure at about $3, $30.37 is what an owner would pay who gave that outcome a 36% chance E (estimate).

Multiply it out. Everything working needs four things: the desmoid trial wins (60% in my base, 80% in my bull), the launch is broad (25% and 40% of wins), polyposis is approved (25% and 50%), and the big-cancer program works (7% and 20%). That is 0.3% in my base and 3.2% in my bull E (estimate). The factors are not independent, so the true figure is higher. I do not think it is 36%.

What a 10x would take. Ten times $30.37 is $304, a market value of $42B on today’s share count of 137.6M. At 15 times earnings that needs about $2.8B of net income, and at a 35% net margin about $8B of annual sales E (estimate). My base case peaks at $0.84B for the desmoid drug and my bull at about $2.2B for the desmoid drug alone. Even the outcome where everything works at my bull sizes is worth about $78 a share, 2.6 times the price and not ten. A tenfold price needs a success bigger than any case I wrote down, within about ten years, which is a 26% gain a year for a decade. I did not measure how often small drug developers have done this and won’t quote a base rate. It is not impossible for a company this size, and it is a long way outside my cases.

Where I come down

Three things moved me: the speed of response, the lack of ovarian damage, and the size of RA Capital’s cheque. The size of the evidence did not: 18 patients at 32 weeks, no length of response and no control group. A response that lasts a year, a clean Phase 3 win, or a first controlled sign of activity in colon or liver cancer would raise my values. Until then the price sits above my bull case. The market may be right, because it weighs what I can’t see: the full data, the FDA talks and how the drug behaves in other tumours.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• Safety from the mechanism. Across all tumour types, severe events rose with dose, and more so in colon cancer patients. In the desmoid group one patient had a grade 4 rise in a muscle enzyme that cleared in seven days. In the larger group two patients on the highest doses had grade 3 adrenal events that paused treatment. No deaths from treatment were reported. The company manages bone and adrenal effects with extra drugs.⁴
• One drug. Nearly everything on this page depends on zolucatetide. Two prostate programs and a beta-catenin degrader have not been in a person.⁴
• Patents. No patent had been issued on the lead drug by March. The Harvard license can be ended if the company misses development milestones and does not fix it.⁴
• Making it. The company owns no plant and relies on single suppliers for some materials. It is moving to a different production method.⁴
• The pen. The Phase 3 trial uses the infusion. The pen may need bridging studies, a human-use test and possibly a separate Phase 2 trial.⁴
• Concentrated holders. RA Capital holds about 23.6%, and on my reading 21.4M of its shares are not locked up, so one seller could move the stock. A lock-up on about 85M other shares ends on Monday 7 December.⁶⁴
• Volatility. Daily moves of 6% or more on about one trading day in five, and a 45% fall in four weeks that I can’t explain, make the price a poor guide to value in either direction.¹
• Disclosure. As an emerging growth company it gives less than most. I had two years of audited figures and a short set of notes.⁴