Setup

Erasca (ERAS, ≈$2.0B mkt cap) is a pre-commercial pan-RAS oncology biotech. The stock fell from $19 (April 21, 2026) to $10.37 (May 15) on a cluster of negative events: (1) Phase 1 data April 27 with mixed signal — 62% uORR8wk in 2L+ KRAS G12X NSCLC plus a Grade 5 fatal pneumonitis at 24mg; (2) RVMD cease-and-desist letter April 24 alleging patent infringement and trade secret misappropriation via the Joyo Pharmatech licensee chain; (3) 10-Q filed May 11 disclosed both for the first time. We're looking at it because the 10-Q + RVMD's parallel filing + the ASCO 2026 plenary May 31 (RVMD daraxonrasib LBA5) together reframe what's actually being adjudicated — and a widely-held bull misread on the Merck KEYTRUDA collaboration.

What the filing says

Cash and runway. $408.5M cash + securities at March 31, 2026 (up from $341.9M at Dec 31, 2025). Funded into H2 2028 per management. No going concern. Operating cash burn ≈$28M/quarter ex the one-time $150M Joyo IPR&D charge. Net loss of $183.4M Q1 2026 vs $31M Q1 2025 is the Joyo charge, not run-rate burn.

Joyo China expansion. ERAS paid Joyo Pharmatech $150M in March 2026 to extend ERAS-0015 territory to mainland China, Hong Kong, and Macau. Trigger: Joyo dosed first patient in JYP0015M101 Phase 2 trial. ERAS chose the more expensive option ($150M post-Phase 2 dosing vs $50M pre-Phase 2). Material because the payment was made BEFORE Phase 1 monotherapy data dropped April 27 — management committed capital before validation — and because RVMD's trade secret claim traces through the Joyo licensee chain.

RVMD IP dispute — primary text. New risk factor (not in March 12, 2026 10-K). Three claims: (1) ERAS-0015 infringes RVMD U.S. Patent No. 12,409,225 under doctrine of equivalents; (2) third party misappropriated RVMD trade secrets in connection with a patent relating to ERAS-0015, ERAS allegedly liable as licensee; (3) "improperly comparative" preclinical disclosures. RVMD demanded ERAS cease all making/using/selling/importation of ERAS-0015 outside Hatch-Waxman safe harbor. ERAS response: claims "without merit," will defend vigorously. Critical new language: "We believe there are multiple potential pathways to address such outcomes, including licenses or other strategic alternatives." Item 1 Legal Proceedings: not currently a party to any material proceedings.

Pipeline acceleration. Pembrolizumab combination cohort opened Q1 2026 (ahead of guidance); monotherapy expansion Q2 2026 (ahead of guidance). H1 2027 combo data confirmed (delayed from H2 2026 in prior disclosure). Subsequent event: Merck KEYTRUDA CTCSA (May 2026) — Merck provides pembrolizumab at no cost; ERAS sponsors and funds. Tango Therapeutics CTCSA (March 2026) — ERAS supplies ERAS-0015 for Tango's vopimetostat (PRMT5) combo. ERAS-4001 BOREALIS-1 Phase 1 monotherapy data H2 2026 unchanged; spend up 66% YoY.

Insider behavior. No officer or director adopted, modified, or terminated any 10b5-1 plan in Q1 2026 — despite the stock running from ≈$2 to ≈$18 over the period. Either confidence or blackout-period caution.

What the market thinks

Stock at $10.37 implies ≈$2.0B market cap, ≈$1.6B enterprise value for the pan-RAS franchise after subtracting net cash. Sum-of-parts intrinsic at our risk-adjusted base case is $3.10-$4.50/share: cash floor $1.31, ERAS-0015 risk-adjusted NPV $1.00-1.60 (25% PoS, $1.5B peak, 10% all-in royalty stack including potential RVMD settlement), ERAS-4001 risk-adjusted NPV $0.30-0.60, M&A optionality $0.50-1.00. The current price implies the market is using higher PoS (40-50%) and/or higher peak sales, OR underweighting the IP overhang and competitive marginalization. Even after the 46% crash, valuation sits well above bear-case intrinsic.

Short interest 11.6%. RSI 17.4 (deeply oversold). MOM1M -43%. Idio variance ≈108% over recent 60d window — the move is dominated by company-specific shocks, not sector or market.

On the IP dispute specifically, market coverage frames it as roughly binary (litigation vs status quo) with formal litigation probability around 35%. Our distribution: settlement-by-license 30%, status quo overhang 30%, formal litigation 25%, M&A-driven resolution 10%, design-around or invalidation 5%. We've shifted ≈10pp from formal litigation to negotiated settlement, with a probability-weighted factor E[r] of -10% over 240d (vs market-implied ~-12.5%).

Why the gap exists

Asymmetric IP disclosure. RVMD's own 10-Q (filed May 6, five days before ERAS's) gave the dispute one sentence in a generic IP risk factor — no patent name, no trade secret detail, no comparative-statements demand mention. ERAS's 10-Q (May 11) ran a full risk factor with verbatim quotes and the new "strategic alternatives" language. RVMD's Q1 earnings call (May 6, Goldsmith) made zero mentions of Erasca, ERAS-0015, or the C&D letter. The expected pattern is reversed: claimant publicizes to anchor narrative, target minimizes to avoid amplifying signal. Reversed pattern historically associates with negotiated outcomes — claimant is preserving optionality and hedging counterclaim/invalidity risk that they explicitly acknowledge in their own filing ("counterclaims challenging the scope, validity, enforceability or ownership of our intellectual property rights"). Sell-side coverage frames the dispute binary; the asymmetric disclosure reframe is not synthesized.

Merck CTCSA misread. The Merck KEYTRUDA collaboration was widely reported as a positive M&A or selection signal. Merck has 9-10 active KEYTRUDA CTCSAs in 2025-26 (EIKN x3, Flare, BioInvent, MacroGenics, Phanes, plus ERAS), is running its own MK-1084 (KRAS G12C) in Phase 3 with KEYTRUDA in 1L NSCLC — a direct competitor to ERAS-0015 in KRAS G12C — and has no comparable CTCSA with RVMD, which runs daraxonrasib + pembrolizumab combo without one. If the CTCSA were a selection signal favoring ERAS, Merck would also be giving RVMD free pembrolizumab. They aren't. Treat the deal as cost optimization for ERAS, not Merck endorsement.

ASCO plenary asymmetry. RVMD daraxonrasib RASolute-302 has the LBA5 plenary slot at ASCO May 31 — top-tier oncology venue, alongside J&J Erleada (LBA1), LLY Verzenio (LBA2), LLY Retevmo (LBA3), Akeso/SMMT ivonescimab (LBA4). ERAS-0015 dropped Phase 1 data April 27 (between ASCO abstract list release and embargo lift), suggesting no plenary submission or no acceptance. Plenary platform creates a self-reinforcing loop: cheaper capital, narrative consolidation, IP leverage. RVMD raised $2.2B in April (post-data, including a $500M 0.50% convertible due 2033 + $1.6B equity), now sits on ≈$4B cash. RVMD doesn't need to escalate IP — narrative + data + cash position already winning.

Comparator footnote unsoundness. ERAS's Phase 1 release attributed a daraxonrasib comparator to Punekar et al. 2025 in Journal of Thoracic Oncology. The Punekar paper is NSCLC; ERAS used it to anchor a PDAC efficacy comparison while RVMD's PDAC-specific Phase 3 RASolute-302 data (mOS 13.2 vs 6.7 mo, HR 0.40, ITT all-comers) had been released two weeks earlier. Cross-indication comparator + uORR8wk vs confirmed ORR. This corroborates RVMD's "improperly compared" allegation and gives RVMD pre-suit settlement leverage on the comparative-statements claim without filing a complaint.

Risks (ranked by impact)

1. M&A bid (the bear's left tail). Acquisition probability 15-18% by April 21, 2027. Defensive scenarios (BMY, AMGN to neutralize a competitor) more likely than offensive (LLY for pipeline gap-fill), but either resolves IP via successor and pays +30-80% premium. A bid neutralizes the entire bear cluster.
2. Bear thesis is mostly priced. ERAS already fell 46% from peak. RSI 17.4 indicates short-term mean-reversion risk. Squeeze setup loaded (11.6% SI, oversold, longs underwater on the dip). Direct short at $10.37 is largely a corpse.
3. ERAS-4001 BOREALIS-1 binary (independent). H2 2026 Phase 1 monotherapy data on the pan-KRAS molecule. Different mechanism than ERAS-0015. ≈35% probability of ≥25% confirmed ORR. Hit revives standalone narrative; miss reduces ERAS to single-asset bet in hostile IP environment.
4. Merck CTCSA could become a real M&A signal (low probability). Operational today. But Merck has clinical DD access to ERAS-0015 data ahead of broader market. If H1 2027 pembrolizumab combo shows dramatic synergy, the relationship becomes M&A-adjacent.
5. Joyo Pharmatech provenance unknown. Joyo is invisible in public-company disclosure outside ERAS's own filings — zero hits in ≈10,625 transcripts and 1,437 China-oncology evidence items. If former-RVMD scientists were involved in the original ERAS-0015 IP, RVMD's trade secret claim viability strengthens materially. Resolves only via formal litigation discovery or Chinese-language regulatory research.
6. Additional safety signals in expansion cohorts. One Grade 5 pneumonitis death at 24mg in 2L mPDAC. If expansion cohorts (now opening) produce additional Grade 4-5 events, the safety profile becomes a separate problem from IP.

Catalysts

What would change our mind

• ERAS files an 8-K announcing a license, royalty, or covenant-not-to-sue with RVMD with material consideration → settlement state confirmed. Bear thesis weakens but valuation impact depends on royalty rate; royalty stack reduces NPV by 10-20%.
• RVMD files a formal patent infringement complaint in federal district court → litigation state confirmed; ERAS to $6.50-8.00 range; bear thesis intensifies materially.
• ASCO Plenary May 31: RVMD G12-subgroup OS HR > 0.40 (our prediction misses) → competitive marginalization narrative weakens; pan-RAS class isn't a one-horse race. Re-evaluate the bear cluster.
• ERAS announces a chemistry variant ("ERAS-0015B") or files an IPR petition at PTAB on the '225 Patent → design-around or invalidation pathway opens; bear thesis weakens.
• Form 4 filings show insider open-market purchases at current prices → competence/governance signal flips positive.
• Joyo provenance research surfaces former-RVMD scientist involvement → trade secret claim viability strengthens; bear thesis intensifies.
• Acquisition announcement → M&A state confirmed; takeout premium captured; remaining bear analysis irrelevant.

Evidence

Memo LR: 0.85 — mildly bearish, mostly catching up to the market. The cross-ticker work produced real but quantitatively narrow edge (≈3% on top of -12% base case). The novel piece is the IP doorway reframe and the asymmetric disclosure pattern; the bear cluster itself is increasingly consensus.