Summary

Andrew Walker, who discloses that he is long KROS, argues that Keros ($KROS) offers an enormous margin of safety that could still be destroyed by poor capital allocation. At roughly $10 per share, Keros has a ~$400 million market cap against about $750 million, or $18 per share, of cash, before assigning value to its Takeda economics. Yet every dollar spent can become “a dollar that could buy back our stock at half of value.”
The collapse of cibotercept, or KER-012, transformed Keros from a clinical-development story into a corporate-governance story. Wells Fargo raised its target from $88 to $111 on December 11; Keros halted the trial for safety issues on December 12, sending shares from the mid-to-high $60s to an $18 opening and eventually below $10. Walker believes the drug is probably worth zero, though management might find another indication or dosing approach after unblinding the data.
The Takeda license for elritercept, or KER-050, is the crown jewel and might alone justify more than Keros’s market capitalization. Takeda paid $200 million upfront, assumed future development costs, and promised more than $1 billion of potential milestones plus low-double-digit to high-teens royalties. Using Takeda’s stated $2–3 billion peak-sales potential, Walker estimates a roughly $400–600 million risk-adjusted NPV while stressing, “Look, I don’t know,” whether the approaching Phase 3 program succeeds.
Walker assigns little current value to KER-065 despite management’s enthusiasm following its Phase 1 DMD data. Sell-side analysts politely relegated the asset to “future upside” and excluded it from sum-of-the-parts valuations pending more clinical evidence—language Walker translates as, “We don’t believe this drug has any value, but we don’t want to piss off the company.” He concedes that the data remain incomplete and he “could be missing something.”
Keros’s historical cost structure is no longer defensible after losing one major program and licensing another. The company spent roughly $175 million in 2024, including about $55 million of personnel expense and another $10 million for professional fees, facilities, and supplies. Walker calls right-sizing that overhead a “pants on fire, all hands on deck mission,” particularly because Takeda will fund KER-050 and KER-012 has been halted.
The board is better aligned than many zombie-biotech boards, but its composition may fit yesterday’s scientific challenge better than today’s financial one. Two directors come from the largest shareholder, which owns 13.3%; another comes from OrbiMed, which owns roughly 4%; and the CEO owns about 4%, partly through options. Still, Walker argues that Keros is now “a capital allocation,” M&A, and governance situation requiring more shareholder and financial expertise.
Walker wants Keros to cut costs, sell KER-012 and KER-065 to the highest bidders, and ultimately return excess capital rather than pursue a “quixotic quest.” If nobody will bid for or develop the programs, that is the market’s answer; if management believes everyone else is wrong, it should acquire the drugs and raise fresh money separately. His “good girl Penny” analogy captures the shareholder-engagement message: investors should say “leave it” before management puts the proverbial chicken bone—and $750 million—into its mouth.

Deep dive

1. A biotech trading below cash has become a governance bet

Walker discloses that he is long KROS and defines a “zombie biotech” as a company valued well below net cash because the market expects management to turn each dollar into 50 cents through unnecessary R&D and overhead. At that point, “it’s actually not anymore about the science”; management must prove the market wrong or return shareholders’ capital.
The structural problem follows a failed blockbuster candidate: management still controls perhaps $500 million raised for the original drug, while its old options are deeply out of the money. Buying another program gives insiders a “heads I win, tails I don’t lose scenario”—success revives their equity, while failure still supplies years of salaries, bonuses, status, and employment.
Walker rejects any “god-given right” to reinvest that legacy cash. Management should return it, form a new vehicle, and ask investors to fund the next idea at a price they voluntarily accept; sunk capital in the existing shell does not confer permanent capital-allocation authority.
The opportunity is unusually broad because former zombies might have traded near 60% of net cash, whereas many now sit at 20–40%. More importantly, some companies trading around half of cash retain genuine crown-jewel assets—Keros and Sage ($SAGE), where Walker remains long and advocates a sale, are his leading examples.

2. KER-012’s overnight failure erased the old Keros thesis

Keros entered the period with three principal programs: cibotercept/KER-012 in pulmonary hypertension, KER-065 in neuromuscular disease, and elritercept/KER-050. Six months earlier, Walker says, virtually the entire stock conversation would have centered on KER-012.
Sell-side expectations captured the concentration. Jefferies initiated coverage in November with shares around $60 and a $107 target, while Wells Fargo lifted its target from $88 to $111 on December 11, presenting the anticipated 2025 Phase 2 readout as a potential blockbuster catalyst.
One day later, Keros stopped the KER-012 trial after safety issues; it subsequently shut the program down in January. Shares opened around $18 after previously trading in the mid-to-high $60s, then drifted below $10.
Management plans to unblind the data and examine alternative indications or dosing. Walker remains skeptical; a Cantor Fitzgerald critique said it was hard to see a path forward unless the safety events could be explained by a subset or something else.

3. The Takeda license now carries most of the asset value

Roughly a week before the KER-012 halt, Keros licensed KER-050 to Takeda. Takeda effectively assumed the program and future development costs, paying $200 million upfront while offering more than $1 billion of potential milestones and royalties ranging from the low double digits to the high teens. Keros would provide paid support during the transition.
Takeda’s own investor materials described the candidate as potentially best in class and estimated $2–3 billion of peak annual revenue. Walker’s illustrative midpoint—$2.5 billion of sales at a blended 15% royalty—would generate more than $300 million of annual royalties for Keros if approved.
The program is about to start Phase 3, with an approval answer potentially around 2028. Walker starts with roughly 50/50 historical odds, allows they “may be materially better than 50/50” because Takeda performed diligence and wrote a large check, but refuses false precision: “Will it be successful? Look, I don’t know.”
After discounting for failure risk and time, Walker estimates the milestones and royalties at roughly $400–600 million, simplifying his working value to $500 million. That exceeds Keros’s ~$400 million market cap; outright success would make the interest worth multiples more, while Takeda—not Keros—funds the path there.

4. The remaining pipeline does not justify the inherited expense base

KER-065 recently completed Phase 1 work targeting DMD. Management cited a “robust pre-clinical package” and said the results equip it to advance into Phase 2, but Walker found the data “a little meh” and says the buy-side and sell-side reactions he encountered were similarly restrained.
Analysts left KER-065 as unmodeled “future upside” pending more clinical data. Walker reads through the diplomacy but keeps the uncertainty intact: full results are not yet available, the program is early, and “maybe I’m missing something. It’s entirely possible.”
Keros spent about $175 million during 2024, when it was supporting three potentially major programs; roughly $55 million went to personnel and another $10 million to professional fees, facilities, and supplies. That structure was understandable then, when clinical spending supported several apparent blockbuster opportunities.
Today, Takeda covers future KER-050 development costs, KER-012 is halted, and KER-065 is the sole independently advancing candidate. Walker has seen no announced right-sizing four or five months after the reset: “The status quo can’t hold,” because a one-drug company entering Phase 2 cannot casually retain infrastructure built for three programs.

5. Shareholders must tell Keros to “leave it” before cash disappears

Keros has about $560 million of year-end cash plus the $200 million Takeda payment received in February 2025—approximately $750 million, or $18 per share, against a ~$10 stock. Walker therefore sees the royalty interest as upside layered over a company trading only slightly above 50% of cash.
Alignment is comparatively encouraging: two directors come from the largest shareholder, which owns 13.3%; another comes from OrbiMed, which owns roughly 4%; and the CEO owns around 4%, though partly through options. Walker’s concern is that this represents less than half the board, while the remaining directors do not own much stock.
Walker argues that the board’s heavy scientific orientation fit last year’s three-trial setup better than today’s capital-allocation problem. His preferred route is to “wrap this up”: cut overhead drastically and auction KER-012 and KER-065 to organizations with development and commercialization infrastructure. If outsiders will not bid, “that’s your answer”; if management wants to continue anyway, it should buy the programs and raise separate capital rather than gamble shareholders’ $750 million.
Walker’s dog Penny supplies the closing governance metaphor. A watched dog will leave a street-side chicken bone when told; an unwatched dog eventually grabs it, forcing frantic cries of “Leave it, leave it, leave it.” Shareholder engagement is the early warning: Keros must either release more data and prove shareholders wrong or “wind it up.”

A Chinese Executive on Trump’s Tariffs and Companies in Mexico Compass: Meals for the Masses - [Business Breakdowns, EP.211]