Episode 191 - July 31, 2026

Summary

• Biotech is decisively outperforming, and the hosts think the rally has legs into year-end. The XBI is +24% YTD versus +9% for the S&P 500 and +6% for health care broadly, even after today’s 3.5% drop pushed it back under $150 from a July 9 high near $164. Graig’s three pillars: FDA headwinds that he believes “have now turned into tailwinds,” 14 U.S. biotech IPOs YTD versus eight in all of last year, and $300B+ of pharma revenue at LOE risk forcing M&A; Tim adds that the XBI tends to drift down without a takeout and would not be surprised to see it finish the year closer to $175–$200.
• Reverse mergers have become as important as the traditional IPO—roughly 10 this year by Tim’s count—and the old stigma has waned. Fund managers cite confidential diligence in the PIPE, speed, and lower cost, plus urgency because “no one wants to wait around and find out” what the midterms bring. A panelist notes syndicates are now “the who’s who” and post-deal performance has been strong, with Yaron flagging the slow close as the one “Achilles’ heel.”
• Tim’s AI thesis: Anthropic’s Claude for Science push matters because agentic leverage for scientists—not molecule design—attacks what actually costs money. “Making molecules with AI doesn’t necessarily solve our industry’s bottleneck,” since the real cost sits in the clinic; tasks that took scientists weeks now take 10 minutes. A panelist’s counter-frame is that AI platforms may have “unlimited power and limited ability” to reduce ideas to practice; AI will be “an enabler” embedded in everyone’s research, not the sole driver, and Graig notes Recursion and Schrödinger charts “haven’t really been all that great.”
• In vivo CAR-T stays white-hot—J&J agreed to total initial payments of $785M for privately held Sail Biomedicines, with a $2.6B acquisition opt-in—and Legend Biotech is under pressure. Yaron highlights Sail’s programmable RNA and circular-RNA eRNA platform. CEO Ying Huang is stepping down, Gilead’s anito-cel is expected to get approval in December and does not seem to cause the same adverse events as Carvykti, and the J&J myeloma opt-in remains unresolved. With two in vivo platform deals done, Yaron sees “even less of a reason to buy their partner.”
• argenx’s $2.2B Forte takeout—believed to be its first acquisition—is Paul’s template for how mini-large caps should avoid the Alexion trap. Companies built on one or two franchises become “a victim of their own success” when the what’s-next question looms (Alexion, Alnylam, BioMarin); buying Forte’s CD122 asset with positive Phase 1B celiac and vitiligo data, ahead of Phase 2B celiac and Phase 1B alopecia areata data, is proactive external innovation. Tradeable kicker: acquirer stocks are rarely down and sometimes up (AbbVie/Apogee), “probably a tailwind for the space too.”
• Lilly’s ataiBeckley play-by-play—from an unnamed pharma’s $125M co-development/co-commercialization offer to $6.75 upfront plus a CVR sweetened from $1.00 to $2.50—is validation that could open the psychedelic floodgates. Paul calls Lilly “probably the most powerful brand in the pharmaceutical space” to put its stamp of approval on DMT; Graig sees BD FOMO kicking in, though Delix Therapeutics’ $5–6B valuation would be “a pretty hefty M&A price” versus Compass and GH Research at $1–2B.
• Adcoms are back, and Brian’s read is that this FDA “was not pulling punches.” Capricor’s deramiocel was opposed 9–3 on protocol-change and multiplicity grounds while Replimune’s RP1 won 10–3; Paul notes UniQure trades on the “FDA vibes thermometer” day by day. Brian’s sizing caveat is that on-the-bubble applications matter less to the sector than drug-pricing reform or interest rates. Dyne also received FDA clearance for a Phase 1 FSHD IND, while Tim described the politicized Fauci hearing as a “tempest in a teapot.”
• The TTR trade resolved ugly: Alnylam cut Amvuttra guidance just after issuing it at JPM, its second miss in three quarters, and the stock fell about 30%. Brian calls the stabilizer-vs.-silencer cross-trial comparison “the dumbest thing ever” because of delayed treatment starts; the discussion framed ESC as at best hypothesis-generating. Other panelists said the drugs may look identical, making the cheaper oral the likely starting point, while separate mea culpas questioned KOL extrapolation and noted that early HELIOS-B combo-versus-mono data had foreshadowed the larger result.
• Today’s read-throughs were harsh: Novo’s ZEUS anti-IL-6 MACE miss hit the NLRP3 cohort—Monte Rosa -61%, BioAge -31%, Neumora -9%—while MapLight undershot Cobenfy. Graig says Neumora’s highly brain-penetrant oral NLRP3 program targets obesity, not cardiovascular outcomes, and a new 13-week rat tox study found no adverse events; data are expected early next year. MapLight’s M1/M4 schizophrenia study missed Cobenfy-level efficacy and its QD formulation was not statistically significant, though tolerability and retention were better.

Deep dive

1. The tape: XBI +24% YTD, and the rally rests on three pillars

• Graig’s setup from Mizuho’s just-published Q2 sector preview: the XBI is back under $150 after touching about $164 on July 9, down 3.5% on the day—“a bit choppy of late”—but fundamentals are “very much intact,” with the index +24% YTD versus +9% for the S&P 500, +8% for the Nasdaq Composite, +6% for health care, and up about 1,600 basis points in three months.
• His three drivers of continued outperformance: FDA headwinds that, after high-profile senior turnover, he believes have “at least turned into tailwinds”; an IPO “locomotive running full steam ahead” (14 U.S. biotech IPOs YTD versus eight all of last year); and M&A propelled by more than $300B in revenue at risk from patent expirations or losses of exclusivity over the next decade.
• The IPO calendar remains active: Sionna Therapeutics debuted last week, Apnimed was expected to begin trading that day, and Tenaya Therapeutics was next, focused on hypertrophic cardiomyopathy and related conditions.
• Tim’s tape-reading heuristic: in weeks without M&A, “the XBI tends to drift down… and the moment someone gets bought, the XBI starts heading in the other direction.” He would not be surprised to see it finish the year closer to $175–$200 than its then-current sub-$150 level.

2. Reverse mergers have quietly matched the IPO as the go-public route

• By Tim’s count, roughly 10 companies have gone public via reverse merger this year—“pretty much as important, if not more important, this year than the traditional IPO, which is quite the change.” Fund managers tell him the PIPE structure allows confidential diligence and a faster listing, while midterm anxiety compresses timelines: “It’s not that people are so pessimistic or optimistic… no one wants to wait around and find out.”
• Another panelist’s observation on stigma decay: companies used to go public this way and remain unknown for two years, but “now it feels like the process has been kind of perfected”—there is no apparent sacrifice in syndicate quality (“it’s often the who’s who”), the deals are said to be cheaper, and post-market performance has been strong, with oversubscribed transactions trading well.
• Yaron, who has worked on several, says speed is “what’s critical” for both investors and management. The one Achilles’ heel is that “it does take time for the deal to close.”

3. Anthropic’s Claude for Science: leverage for scientists, not a molecule machine

• Tim’s core argument, prompted by speculation that Anthropic “could just snap up Bristol Myers or AbbVie”: AI-for-molecules misses the point because most of the cost—the real cost—is on the clinical side, where AI is helpful but less helpful. The real transformation is agentic leverage: protocols, experiment design, write-ups, article review, and revision—“things that used to take scientists weeks and months that you can do in 10 minutes now.”
• K-Dense, for example, added over 150,000 scientists to its site in the last couple of months. Tim sees the space as “red-hot” and AI as potentially transformational to the process of doing science, not merely to molecule generation.
• A panelist’s tempering view, informed by covering informatics-designed drugs such as Relay, is that the platform can have “unlimited power and limited ability” to reduce ideas to practice at acceptable cost, given inevitable failures. AI “is going to be an enabler,” embedded in everyone’s research, while everything around it may be equally or more critical.
• Graig’s scoreboard check: AI drug discovery has been tried before, and levered names Recursion and Schrödinger have stock charts that “haven’t really been all that great”—though he thinks “we’re in a newer age now.”

4. In vivo CAR-T stays white-hot—and Legend Biotech is under pressure

• The deal: J&J is making total initial payments of $785M to license Sail Biomedicines’ novel in vivo CAR-T platform and can opt to acquire Sail for another $2.6B. Graig said he thought he had read that Sail was a Flagship Pioneering company, but the transcript does not establish that point definitively.
• Yaron’s technology case: Sail uses programmable RNA therapeutics delivered through chemically encoded nanoparticles, moving away from PEG toward stabilizing lipids. The approach is not a standard lipid nanoparticle and is not limited to targeting hepatocytes. Its eRNA uses a circular RNA construct that is more durable than linear mRNA, which “maxes out within a day”; the cargo is fairly large and has broad transduction potential.
• Graig’s comparables are Lilly’s $2.4B Orna deal for an in vivo CAR-T platform in I&I and Lilly’s separate Kelonia Therapeutics deal for in vivo CAR-T in hematology, which he cited at $7B.
• Legend is at a crossroads. CEO Ying Huang announced Monday that he is stepping down after accepting a presumably CEO role at a new venture-backed oncology company. Gilead’s anito-cel is expected to get approval in December and does not seem to cause the same immune-related adverse events as Carvykti—parkinsonism, Bell’s palsy, colitis, and cytokine release syndrome—though it might be “a smidgen less potent.”
• A second uncertainty is whether J&J will opt in to develop in vivo CAR-T for myeloma; if it does not, neither company can enter myeloma without the other. A third is how Legend will broadly develop its in vivo platform despite good lymphoma data. Yaron has argued that a J&J acquisition of Legend is unlikely, and after J&J’s two broad in vivo platform deals there is “even less of a reason to buy their partner.”

5. argenx buys Forte: how mini-large caps should spend success

• Yaron covers both sides of argenx’s $2.2B Forte acquisition, which he believes is argenx’s first acquisition, under new CEO Karen Massey. The company is committing to external innovation despite a strong internal pipeline and Vyvgart’s continued performance; Vyvgart has beaten for roughly its 14th or 15th consecutive quarter since launch.
• Forte’s key asset targets the CD122 receptor, where IL-15 and IL-2 signal. It has produced positive Phase 1B data in celiac disease and vitiligo. Argenx was awaiting confirmatory randomized Phase 2B celiac data and Phase 1B alopecia areata data by year-end. Yaron is optimistic about the asset.
• The competitive frame includes Teva, with Royalty Pharma, developing an IL-15 antibody that binds the cytokine rather than the receptor; it has positive vitiligo data, with celiac data expected in the second half. First Tracks also has a CD122 antibody expected to read out Phase 1A/Phase 1B celiac data in the second half.
• The deal was “a little telegraphed”: argenx invested in Forte’s offering in late April or early May, Forte pulled its celiac-data guidance forward to “soon” or “imminently,” and argenx reiterated its desire to do M&A on Thursday. Forte “absolutely ripped” Friday and was acquired Monday.
• Paul’s broader thesis is that companies in the “mini-large-cap range” built on one or two franchises can become “a little bit of a victim of their own success.” Alexion is the cautionary tale: Soliris and Ultomiris were highly successful, but an early- to mid-stage competitor created an existential threat to terminal value and exposed the need to backfill the pipeline. Alnylam, Insmed, and BioMarin face versions of the “what’s next?” question, while Vertex spent years in limbo before pipeline assets broke through.
• When a franchise such as Vyvgart or mavacamten is performing well, “it’s easy to be risk-averse.” But market dynamics, investor perception, and multiples can change quickly, often because of factors outside a company’s control. Acquirer stocks have rarely gone down lately and sometimes rise, as in the AbbVie/Apogee situation—“probably a tailwind for the space too.”

6. Lilly’s ataiBeckley play-by-play opens the psychedelic floodgates

• Graig walked through the SEC-filing history: an unnamed pharmaceutical company approached ataiBeckley in December—one month after its merger with Beckley Psytech—offering $125M upfront for a co-development and co-commercialization deal around BPL-003 for treatment-resistant depression.
• Lilly joined in early June with an initial deal, then sweetened its offer on June 11 to $6.75 upfront and added a $1-per-share CVR tied to potential FDA approval of the Phase 2 asset VLS-01. After further back-and-forth, Lilly raised the CVR to $2.50, which completed the deal.
• Paul’s validation argument is that the key question was whether a traditional pharmaceutical company concerned about perception would buy a psychedelics company. Lilly is now “probably the most powerful brand in the pharmaceutical space” putting “its stamp of approval on DMT,” a compound with “tremendous efficacy data” but an experience that is “pretty out there” from the psychedelic perspective. He also sees Lilly’s infrastructure-building as helpful to the space, much as J&J’s work with Spravato was.
• Graig sees an element of BD FOMO: when one company moves, others may worry about being left behind. He noted that Delix Therapeutics, after strong data, was valued at roughly $5–6B—a “pretty hefty M&A price”—while Compass Pathways and GH Research had lower public valuations of roughly $1–2B.

7. Adcoms are back—and this FDA “was not pulling punches”

• Brian’s beat: two “really good” advisory committees this week for controversial applications—Capricor’s deramiocel in DMD and Replimune’s RP1 in melanoma. Briefing documents hit both stocks days early, “in classic Wall Street fashion,” as the FDA criticized the data.
• Capricor’s committee voted 9–3 against, focusing on statistics and how protocol changes can bias results. Brian’s concern “really resonates” with him; his Twitter profile features a multiplicity calculation for Type I error. Replimune’s discussion centered more on measuring intratumoral injection activity versus using RECIST for systemic therapy, and the vote was 10–3 in favor.
• Brian’s larger point is that advisory committees make scientific discourse and the search for consensus visible when clinical information is imperfect. He criticized the prior decision under Marty Makary and Vinay Prasad to stop holding these meetings and internalize the process. But the panels also showed that their departures did not mean the FDA would give everyone the go-ahead: “the FDA was not pulling punches at either of these advisory committees.”
• Paul notes that UniQure trades around the “FDA vibes thermometer”—down on the Capricor documents and up after Replimune. Brian’s sizing caveat is that on-the-bubble applications are a small part of the sector and less consequential than drug-pricing reform or interest rates.
• Dyne received FDA clearance for its IND application for a Phase 1 FSHD study, a positive step for a program that has been slow to advance and that parallels Avidity’s FSHD work, which is ahead after Novartis acquired Avidity.
• Tim described Anthony Fauci’s subpoenaed appearance before the Senate Homeland Security and Governmental Affairs Committee, chaired by Rand Paul. Fauci invoked the Fifth Amendment more than 100 times over roughly three hours, while Paul released more than 1,000 pages of Fauci’s 2019–2022 personal diary, which contained no evidence about COVID’s origin. Tim called the event a politicized “tempest in a teapot”; Yaron said it was unfortunate grandstanding that could further erode public trust in health agencies.

8. Alnylam’s guide cut and the TTR endgame: the cheaper oral wins

• Paul described Alnylam’s roughly 30% sell-off after the company lowered Amvuttra guidance on its Q2 call, just after issuing it at J.P. Morgan. It was the second consensus miss in three quarters, creating both fundamental and emotional frustration, though the revised guide still implied more than 50% growth in the second half of 2026 versus the second half of 2025.
• Ionis’s TTR outcome study for its silencer failed, seemingly because it showed no benefit on top of tafamidis. Alnylam’s next-generation silencer, which could nearly double the profitability of its TTR franchise by eliminating royalties, is in an outcomes study that is almost functionally being run as a tafamidis combination study.
• Yaron’s ESC preview: Wainua monotherapy versus placebo showed a 0.71 hazard ratio, the closest apples-to-apples comparison with Amvuttra’s HELIOS-B data. Ionis believes the totality of the data, including an independent meta-analysis, will show Amvuttra and Wainua are “very similar to—slash, identical to.” It also believes the stabilizers performed better than historically, potentially because more patients were NYHA Class I.
• Brian called the stabilizer-versus-silencer comparison “the dumbest thing ever” from a context perspective: stabilizer patients were treated much longer, anywhere from six months to years, making the comparison resemble a delayed-start analysis. He stressed that this was not a criticism of Ionis’s drug developers, whom he called “legends.”
• The discussion characterized ESC as, at best, hypothesis-generating for Alnylam and suggested the company may need to change its outcomes study to restore confidence. One panelist said the Ionis train in TTR had “absolutely left the station” and that the data were now mainly relevant to Alnylam and BridgeBio read-throughs. Another panelist said that if the drugs look identical, the practical question is whether prescribers start with a cheaper oral or an expensive injectable; “most people believe they’ll start with a cheaper oral.”
• Other panelists questioned whether the KOL sample favoring silencers after HELIOS-B was extrapolatable to busy community physicians, who may choose what is easiest. Another panelist said the original underpowered HELIOS-B combo-versus-mono result already showed no benefit, was dismissed as statistically irrelevant, and ultimately proved clinically correct in the larger sample.
• A further observation was that the mortality reduction was very large while functional endpoints such as KCCQ and the six-minute walk were less convincing, potentially reflecting the importance of the treatment context, population, and disease severity.

9. Today’s wreckage: Novo’s ZEUS miss torches NLRP3; MapLight undershoots Cobenfy

• Novo Nordisk shares fell almost 10% after its anti-IL-6 ligand antibody missed the primary MACE endpoint in the Phase 3 ZEUS study. The result undermined the idea that lowering hs-CRP would necessarily improve cardiovascular outcomes and generated negative read-throughs for NLRP3 inhibitor developers: Monte Rosa -61%, BioAge -31%, and Neumora -9% intraday.
• Graig’s defense of Neumora, which he covers: its oral NLRP3 inhibitor is highly brain-penetrant and aimed at obesity rather than cardiovascular outcomes. After concern about possible adverse events, the company ran a separate new 13-week rat toxicology study with no adverse events and plans to move forward; data are expected early next year. Neumora also has an M4-PAM muscarinic franchise with updates expected later this year.
• MapLight’s M1/M4 schizophrenia data did not show Cobenfy-level efficacy, and the QD formulation—hoped to be a differentiator—did not show statistically significant efficacy. Tolerability and retention were somewhat better, but the stock sold off significantly before recovering some.
• Paul’s broader CNS point is that effect size is “not the best correlate of commercial success in almost any neuroscience category.” Outside Ocrevus in MS, he pointed to psychiatry and epilepsy, and even to the pre-Ocrevus MS market, as examples where commercial success was not simply efficacy-driven.
• The next question for MapLight is Alzheimer’s disease psychosis, where its dosing may have more room to differentiate from Cobenfy. Paul is cautiously optimistic that muscarinics may work in ADP, while noting that many antipsychotics probably work but are limited by the black-box warning for increased mortality. Cobenfy’s ADP data and MapLight’s data were both pushed to next year.