Newsletter · 2026-09-11 · Ashutosh Agarwal

Three Novartis Failures Wipe Out 40 Billion Dollars in a Week - The Biotech Patent Cliff & M&A - Week of September 11, 2026

The Biotech Patent Cliff & M&A for the week of September 11, 2026: three separate Novartis setbacks erased more than $40 billion of market value in eight days and dragged Amgen and Lilly down with them, EY data showed big pharma's share of biopharma spend falling to 45% as mid-cap buyers crowd in, Lilly closed its AtaiBeckley bolt-on, and CMS said 90% of branded drugs are now under most favored nation pricing ahead of the September 29 tariff cliff.

The Biotech Patent Cliff & M&A

Week of September 11, 2026: Three Novartis Failures Wipe Out 40 Billion Dollars in a Week

TL;DR

• Novartis had a catastrophic week, more than $40 billion of market value gone in eight days, after three separate failures: three patient deaths in its autoimmune CAR-T program, a big heart-drug flop (pelacarsen), and a Phase 3 miss (Harbor) on the very asset it bought for roughly $12 billion when it took out Avidity Biosciences earlier this year. It is the first real reminder that the deal supercycle is buying hope, not certainty.
• The M&A machine took a breather on the mega-deals but is broadening out. No new $1B+ takeover from the big buyers this week, but the structural picture per EY: only five deals topped $10B in the first half, big pharma's share of spend fell to 45% (from 69% in 2025), and mid-cap and specialty buyers are crowding in. Lilly closed its next bolt-on (AtaiBeckley) anyway.
• The drug-pricing screw keeps turning. CMS chief Dr. Oz says 90% of branded drugs sold in America are now under "most favored nation" pricing and touts GLP-1s at $50/month for some seniors. The Section 232 tariff cliff still lands September 29, and the mandatory Medicare pricing models (GLOBE/GUARD) look set to slip to January.

What's new

1. Novartis's $40 billion week, and why every dealmaker should be nervous

The single biggest event of the week didn't come from a company on our coverage list, but it lands squarely on the thesis. On BioCentury This Week, Ep. 386 (Sept. 9), the editorial team, Simone Fishburn, Steve Usdin, Lauren Martz, and Stephen Hansen, walked through what Hansen called a "bad run" that wiped more than $40 billion off Novartis's market cap in a little over a week (BioCentury This Week). These are journalists and analysts, not company insiders, but their reporting is the clearest account out there. Three things went wrong, back to back:

• Three deaths in the RAP-cel (rapcabtagene autoleucel) CAR-T cell therapy trial for autoimmune disease.
• The Horizon Phase 3 outcomes trial for pelacarsen, an antisense drug that lowers Lp(a), a genetic risk marker for heart attacks, failed. As Fishburn put it, this was "probably the biggest study that everyone was looking forward to this year."
• The Harbor Phase 3 trial in myotonic dystrophy (DM1) failed, and this was the lead asset Novartis acquired when it bought Avidity Biosciences for roughly $12 billion earlier in the year.

Why it moves numbers: the pelacarsen failure is the one that hurts, because it drags in others. Novartis lost most of the $40B on that single readout. Amgen fell about 9% (roughly $21 billion) on the day because it has a rival Lp(a) drug (olpasiran) reading out on a similar heart-attack-survivor population, and Eli Lilly ticked down 2% (roughly $25 billion) on the same read-across. The debate Martz laid out: was it the molecule, the target, or the trial design? Her key technical point is worth holding onto: antisense drugs like pelacarsen knock down Lp(a) by "up to about 80%," while the newer siRNA drugs (Amgen, Arrowhead) get "close to 100%, in the upper 90s." If the siRNA version works where the antisense one failed, she said, "that would have a very big impact on the whole field of antisense versus siRNA." Tellingly, Martz noted Novartis itself had been hedging: it paid $160 million and $185 million upfront in two separate 2024 to 2025 deals with China's Argo Biopharma to license siRNA drugs for cardiovascular disease, one of which (DII-235) targets Lp(a) and reads out next year.

The bigger lesson for this newsletter: when you pay a mega-premium for a single hot asset (Avidity's $12B), you own the binary. Novartis just found out the hard way.

2. The M&A picture: fewer whales, more minnows, and a China surge

Two podcasts this week gave the best independent read on where the deal wave actually is. On BioTalk with Rich Bendis (Sept. 8), EY's Subin Baral, a life-sciences deals advisor, so an industry insider on the transaction side, laid out the hard numbers (BioTalk with Rich Bendis):

• The "large majority" of first-half deals were in the $1 to 10 billion range, and only five deals exceeded $10 billion in all of H1 2026.
• Big pharma's share of biopharma spend fell to 45% in H1 2026, from 69% in full-year 2025. In plain terms: the buyer pool is broadening, mid-sized and specialty pharma are now doing deals to scale up or grab differentiated technology, not just the giants.
• The bar for what gets bought has risen sharply: "the me-too type investments... are no longer as attractive." Buyers want differentiated science with data behind it. Oncology is still the biggest slice of spend; the hot modalities are cell and gene therapy, ADCs (antibody-drug conjugates), and bispecific antibodies, with immunology and neuroscience rising.
• China is now roughly half the action: Baral said about 54% of total US and European alliances in H1 went to China (including "biobucks"/milestone-heavy structures), versus 34% for the prior full year.
• His forecast: an "accelerated" next 12 to 18 months, "2026 to be the year," with competition for differentiated assets intensifying and valuation premiums staying "very, very high." With VC and IPO financing still tight, cash-strapped biotechs will be "forced" to the deal table.

That maps almost exactly onto BioCentury's fourth annual pharma-deals analysis, presented by Lauren Martz on the same Ep. 386. She counted 221 deals across 21 top pharma companies over the trailing 12 months, activity up versus last year and nearly back to two years ago. The theme: pharma is chasing "disruptive potential." Bispecifics are the modality rising fastest, especially T-cell engagers (and even myeloid-cell engagers) aimed at resetting the immune system in autoimmune disease, the same goal as a CAR-T but more accessible. In-vivo CAR-Ts and siRNA deals were also way up; there were essentially no antisense deals in the set, more evidence the field is voting for siRNA. And the standout buyer, again, was Eli Lilly, which signed far more deals than anyone, more than double its nearest peers (GSK, Novartis, and Sanofi, tied for second). (Recall from last week: that streak now runs to Lilly's 13th takeout of 2026, with $31.5B+ of M&A committed year-to-date.)

3. Lilly closes AtaiBeckley; the bolt-ons keep coming

Consistent with all of the above, Lilly completed its acquisition of AtaiBeckley (ATAI) on Sept. 11 (per The Fly). AtaiBeckley is a clinical-stage company developing rapid-acting "neuroplastogens" for mental health; the lead program, BPL-003, targets treatment-resistant depression. Reported terms (verify before trading on them, sourced from web, not a wire permalink): $6.75/share cash upfront (roughly $2.8B) plus a $2.50/share contingent value right (CVR) that could bring total value to roughly $3.8B. This is the textbook mid-cycle structure Baral and Martz described: a differentiated, higher-risk asset, bought with a CVR to bridge the valuation gap. It is also Lilly quietly extending its lead as the industry's most relentless acquirer.

4. AstraZeneca: a landmark approval, then a same-week miss

Our lone cliff-exposed name with real newsflow had a genuinely split week. On Sept. 4, the FDA granted accelerated approval to camizestrant (Etcamah) in combination with a CDK4/6 inhibitor for HR+/HER2- advanced breast cancer once an ESR1 mutation is detected, the first cancer approval based on detecting a resistance mutation in the blood (ctDNA) before the disease visibly progresses on a scan (FDA). In the SERENA-6 trial, switching to camizestrant extended median progression-free survival to 16 months vs. 9.2 months, a 56% reduction in the risk of progression or death. Guardant's liquid-biopsy test was approved the same day as the companion diagnostic.

Then, after the close on Sept. 11, AZN said the SERENA-4 trial, camizestrant plus palbociclib in the first-line setting, missed its primary PFS endpoint (numerical improvement only). Shares slipped about 2% to roughly $157 after hours (AstraZeneca). Net-net: the ctDNA-guided switch strategy is real and approved, but the dream of moving camizestrant to broad first-line use just got harder. AZN otherwise had a strong pipeline week (a COPD biologic, tozorakimab, hit its Phase 3 and got Priority Review; Imfinzi plus tarlatamab hit in first-line small-cell lung cancer), which cushions the blow.

5. Policy: "90% of branded drugs" now under MFN, and the tariff clock is ticking

The pricing story got a loud, on-the-record push from the top. On Squawk Pod (Sept. 8), CMS Administrator Dr. Mehmet Oz, the operator running the program, so weigh accordingly, made striking claims (Squawk Pod):

"Now 90% of all the drugs, branded products sold in America are under most favored nation drug pricing."

He called it "the biggest drop in prices in 63 years" (correcting himself that prices are still historically high, but the decline is the largest), said nine more companies signed the week before, and framed it like NATO burden-sharing: the US pays "more than twice as much percentage of our GDP for drug prices" as Europeans, so "they need to go up a little bit. We need to come down." He said the UK has already agreed to raise its prices. On GLP-1s, he claimed some Medicare beneficiaries now get them at $50/month vs. over $1,000 historically, with over 600,000 people joining that Medicare program in the last two months. Crucially, he admitted the deals are voluntary and time-limited, "the contracts will expire after the administration completes its term," which is why the White House wants Congress to codify them.

The nuance came from BioCentury's Steve Usdin (Washington editor, journalist and analyst) on Ep. 386, whose focus is "who wasn't at the White House." Companies like Vertex, Biogen, and Alnylam are under pressure to do MFN deals and, so far, show "no indication that they're going to succumb." He reported the fine print starting to emerge: Incyte signed a quieter MFN deal (disclosed in an SEC filing) giving state Medicaid programs international-reference pricing on Jakafi; BridgeBio secured a "forward-looking exclusion" for orphan-only medicines (its entire pipeline). And he laid out the two "hammers" the administration is using:

• Mandatory Medicare models, GLOBE (Part B) and GUARD (Part D). GLOBE was supposed to start Oct. 1; that's "not happening." Expect GLOBE to slip and both to start around Jan. 1, followed quickly by litigation.
• Section 232 tariffs of up to 100% on certain drugs, scheduled for Sept. 29, the deadline that "brought the foreign companies to the White House." Usdin expects "more companies with tariff exposure announcing MFN deals in the coming weeks before September 29."

And there's a slower-burn angle. On 340B Unscripted, Ep. 99 (Sept. 7), former CMS official Joella Carman (a regulator's-eye view) argued the administration's desire to get manufacturers onto MFN may be shaping how aggressively HRSA polices the 340B discount program, a possible quiet quid pro quo: cooperate on pricing, get "some reward via the 340B program" (340B Unscripted). She noted 22 states have now passed laws restricting manufacturer 340B conditions, and that Eli Lilly has taken the most aggressive stance on contract-pharmacy restrictions, even leaning on the IRA's duplicate-discount language, despite having no IRA-negotiated drugs until 2028.

The debate

The supercycle bull. The setup hasn't changed and this week reinforced it. A roughly $200B-a-decade patent cliff is forcing the big cliff-exposed names to buy growth, and the buyers have the balance sheets to do it. EY's Baral is outright "bullish," calling for an accelerated 12 to 18 months with H1 already outpacing record levels; BioCentury counts deal activity climbing back toward its peak. Financing is still tight for small biotech, so targets are cheap-ish and "forced" to sell. The buyer pool is deeper than ever, not just Lilly and the giants, but a wave of mid-cap and specialty acquirers. If you own quality SMID assets with differentiated, de-risked data, you are holding the exact currency everyone wants.

The cliff-erosion bear. Novartis just showed you the other side of the trade. When you pay a fat premium for a single disruptive asset, you own a coin flip, and $12B (Avidity/Harbor), plus the pelacarsen and CAR-T disasters, evaporated in a week. The hottest modalities everyone is bidding up, in-vivo CAR-T and T-cell engagers for autoimmune, are exactly where the safety and efficacy questions are least answered (three deaths in Novartis's autoimmune CAR-T is not a footnote). Meanwhile the policy vise is real and getting tighter: 90% of branded drugs under MFN, a 100% tariff wall on Sept. 29, mandatory Medicare pricing models coming in January, and litigation guaranteed. The math that justifies the premiums assumes pricing power that Washington is actively dismantling.

The synthesis. Both are right, and that's the point: this is a market that's paying peak premiums for peak uncertainty, and the dispersion of outcomes is widening, not narrowing. Novartis is the tell, the deal wave is not a smooth escalator up, it's a series of binary bets, and we just watched a great operator lose $40B on three of them at once. Baral's framing that execution is now the differentiator ("many deals... haven't realized the value they anticipated") argues for the disciplined serial bolt-on machine (Lilly, structured with CVRs to share the risk) over the swing-for-the-fences single-asset mega-deal. On policy, don't fade it: MFN at "90% of branded drugs" and a real tariff deadline are not noise, and the names refusing to sign (Vertex, Biogen, Alnylam) are carrying a tail risk the market isn't pricing. The cleanest place to be remains high-quality, differentiated SMID assets with data in hand, the thing every buyer says they want and can't manufacture internally.

Stocks in play

Read-throughs

• CAR-T and the autoimmune "reset" trade. The three deaths in Novartis's RAP-cel program are the loudest data point yet in a building safety cloud, it sits right alongside BMY's zola-cel autoimmune pause (since June, per WSJ via MT Newswires) and program-wide holds at Novartis after reported cases of immune-effector-cell-associated hemophagocytic syndrome (IEC-HS). Everyone from Lilly (its in-vivo CAR-T buys, Colonia/Orna) down is paying up for exactly this modality. If the class carries a mortality signal, the premiums paid for in-vivo CAR-T and T-cell engagers in autoimmune, the fastest-rising deal category per BioCentury, need to come down. Watch smaller pure-plays (Kyverna, Cabaletta, Autolus) for sympathy moves.
• Antisense vs. siRNA. BioCentury's read, no antisense deals in the annual set, siRNA deals "way up," plus the pelacarsen failure is a double blow to the antisense franchise (read-through to Ionis) and a relative win for siRNA platforms (Arrowhead, Alnylam). Amgen's olpasiran readout (next roughly 12 months, siRNA, similar population) is now the single most important cardiovascular event on the calendar; a win reframes the whole modality debate.
• SMID takeout sentiment. For the third straight week, there is zero single-name takeout chatter on the podcasts for our SMID list (Summit, Madrigal, Viking, Crinetics, Cytokinetics, Insmed, Vaxcyte, Krystal, Roivant, Revolution Medicines). Moves were data- and sell-side-driven: Summit got a Jefferies upgrade (Buy, PT $25 from $15); Revolution Medicines got a Morgan Stanley Overweight initiation ($255); Roivant's Pulmovant unit popped roughly 20% on a Phase 2 PH-ILD win. The EY and BioCentury message, buyers want differentiated, de-risked assets and premiums are high, is bullish for this cohort in the abstract, but the market is still waiting for the bids.
• China as the sourcing engine. Baral's 54% figure and BioCentury's China-deal surge both point the same way: an ever-larger share of the assets filling the cliff is being licensed from China, not acquired outright (no one is buying Chinese biotechs, "too many hoops," per BioCentury). Watch the Treasury and Congress "Coins Act" discussions Usdin flagged; any move to restrict US-China biotech dealmaking would choke a pipeline the whole supercycle now depends on.
• Bankers and CROs. More deals, smaller average size, more mid-cap buyers, and a premium on post-close execution is a volume tailwind for advisors and a mixed bag for CROs, the shift toward earlier-stage, differentiated assets favors specialized development capacity over commodity capacity.

What changed vs. last week

• The CAR-T "pause" became a body count. Last week we flagged the BMY zola-cel pause as market chatter and a Novartis CAR-T pause alongside it. This week it hardened: Novartis disclosed three deaths in its autoimmune CAR-T program, and BioCentury put it at the center of a $40B market-cap loss. This is now a genuine class-wide safety question, not a rumor.
• Novartis went from "best in class" to cautionary tale. A name barely on the radar last week became the story, the Lp(a) (pelacarsen/Horizon) and DM1 (Harbor/Avidity) failures are new this week.
• AZN camizestrant: approval confirmed, then complicated. Last week the accelerated approval was the win; this week's SERENA-4 1L miss (9/11) is the new wrinkle that caps the upside case.
• MFN got a hard number and a face. Last week MFN was Usdin's "next shoe to drop." This week the CMS Administrator himself put it at "90% of branded drugs," and BioCentury named the holdouts (Vertex, Biogen, Alnylam) and the carve-out mechanics (Incyte, BridgeBio orphan exclusion). The mandatory GLOBE/GUARD models are now expected to slip to January. Section 232's Sept. 29 deadline is unchanged, two weeks out.
• Open items, updated: Sangamo 363, Lilly's platform and prion piece closed around Sept. 4 ($50M); PTC's Fabry piece court-approved but awaiting HSR; the Sept. 10 hearing was cancelled (per web, verify). Novo v. Lilly, still no ruling, decision reserved (and it's a Lanham Act false-advertising suit, not a compounding case, correcting a prior framing). Jazz/Actio, still on track for Q4. SRRK PDUFA, still Sept. 30, on track.
• Still under-covered (silence is signal): For the third straight week, subcutaneous-Keytruda vs. IV-biosimilar defense got no dedicated podcast treatment, still the most important untouched debate for the biggest single cliff exposure (MRK). And MRK, PFE, JNJ (standalone), GILD, VRTX had no substantive management or podcast commentary, big-pharma cliff talk again ran entirely through the deal-and-data lens. On the news side, Merck did draw a wave of sell-side price-target hikes (Wells Fargo $170, Guggenheim $170, Leerink $161 from $127, HSBC $172) reflecting rising confidence it can grow through the Keytruda cliff, a notable vote of confidence even without management on the mic.