Arcus Biosciences stock initiated at Buy by TD Cowen on RCC drug

TD Cowen initiated Arcus Biosciences (RCUS) at Buy, arguing casdatifan could become a leading renal-cell-carcinoma treatment based on 15.1-month median progression-free survival versus 5.6 months for Welireg. Arcus reported Q2 2026 adjusted EPS of -$0.72, ahead of the -$0.90 expected loss, while $41 million in revenue beat the $22.22 million forecast. Upcoming October data could derisk Phase III development in a renal-cell-carcinoma market estimated above $10 billion; Truist raised its target to $40 from $35, while published targets reach $47.

Analysis

RCUS is now an event-driven oncology beta rather than a conventional earnings story: the October dataset must demonstrate that activity is not merely the established contribution of cabozantinib or checkpoint blockade. The quoted progression-free-survival comparison versus MRK’s Welireg is cross-trial and vulnerable to differences in line of therapy, risk mix, follow-up, and assessment methodology; investors should assign little value to it until durability, response depth, discontinuation rates, and dose intensity are disclosed. A clean signal could expand the probability of a registrational path, but the existing valuation already embeds meaningful pipeline success after the prior rerating.

The most relevant competitive read-through is EXEL, whose cabozantinib could remain embedded in a successful second-line regimen; RCUS success therefore need not be negative for EXEL and may modestly extend the franchise’s combination utility. MRK has far greater commercial exposure through Welireg, but RCC substitution alone is unlikely to move MRK earnings or its multiple. The larger second-order risk for RCUS is commercial economics: even strong efficacy may not translate into a premium product if a regimen increases toxicity, monitoring burden, or payer resistance relative to an incumbent oral HIF-2 option.

Near term, the stock can trade materially on abstract-level details before investors can assess mature survival data. Over 1-3 months, a low primary-progression rate, credible overall-survival trend, and tolerable safety profile would support Phase III probability expansion; a miss on any one of these will expose a valuation supported primarily by a single asset. Over 6-18 months, the key question becomes trial design, enrollment speed, and whether the cash runway remains intact without dilutive financing before decisive registrational milestones.

Contrarian view: consensus appears to be treating the reported PFS differential as evidence of intrinsic superiority, while combination trials often create an attribution problem. The more asymmetric setup may be post-data: a positive but non-definitive release could still be sold if safety, censoring, or control-arm comparability are inadequate, whereas a genuinely differentiated monotherapy survival signal would be more valuable than headline PFS alone.

Key Decisions for Investors

• Maintain RCUS as a small event position only; add after the October release only if primary progression is below 20%, median PFS is approximately 15 months or better with mature follow-up, and no unexpected discontinuation signal emerges. Size for a 25-35% downside on ambiguous or negative data versus 50%+ upside on a clearly de-risking dataset.
• Do not use the Welireg cross-trial PFS comparison as a basis for a short MRK position; RCC/HIF-2 displacement is immaterial to MRK at portfolio scale. Reassess only if RCUS presents randomized evidence or guidance indicating material share capture.
• Watch EXEL for a constructive read-through rather than treating it as a casualty: positive casdatifan-plus-cabozantinib efficacy with maintained cabo dosing supports cabozantinib’s combination relevance. A long RCUS/long EXEL basket is preferable to an RCUS-versus-EXEL pair until regimen economics and prescribing behavior are clearer.
• Before considering RCUS calls, verify implied volatility, option liquidity, exact presentation timing, and the full abstract. If pre-event implied volatility exceeds the expected post-data move, wait for data rather than paying for binary exposure.