Merck: Keytruda Dependency and AI's Acceleration of Oncology Competition

Executive Summary

Merck & Co. is the most concentrated single-asset large-cap pharmaceutical company in the world. Keytruda (pembrolizumab), its anti-PD-1 immunotherapy, generated approximately $25 billion in 2024 — nearly 43% of total company revenue of approximately $63 billion. With Keytruda's core U.S. composition-of-matter patents expiring around 2028 and biosimilar entry anticipated in the 2028–2030 window, Merck faces a revenue cliff that makes Abbvie's Humira transition look manageable by comparison. AI is not the primary driver of this cliff, but it is a powerful accelerant of competitive dynamics in immuno-oncology that will determine how quickly Keytruda's replacement therapies arrive from competitors.

Business Through an AI Lens

Merck's portfolio beyond Keytruda includes vaccines (Gardasil, Varivax), animal health (Banfield, Bravecto), cardiometabolic (Winrevair, islatravir for HIV), and a pipeline across oncology combinations, ADCs (antibody-drug conjugates), and small molecule engagers. The company has made significant AI investments: its Merck Digital Science Studio deploys machine learning for target identification, molecular dynamics simulation, and clinical trial optimization. Partnerships with Absci (AI-designed antibodies) and Linus Health (AI-powered cognitive assessment) expand the AI footprint.

In oncology, AI's most significant impact is accelerating the development of PD-1/PD-L1 successor therapies. LAG-3 inhibitors, TIGIT inhibitors, CTLA-4 next-generation antibodies, bispecific checkpoint inhibitors, and wholly novel mechanisms (PVRIG, LILRB2) are all being developed with AI assistance for target selection, lead optimization, and combination therapy design. The pace of oncology pipeline development has accelerated by an estimated 30–40% over the past five years, with AI tools being a primary driver.

For Merck specifically, this creates a paradox: Keytruda is itself an AI-era therapy in terms of biomarker stratification (TMB-H, MSI-H patient selection), but the same AI tools that enabled its clinical success are now being used by competitors to design therapies that outperform it in specific tumor microenvironments.

Revenue Exposure

Gardasil deserves separate attention. Its 2024 revenue declined significantly due to China distribution disruptions (China had been approximately 30–35% of Gardasil global revenue). Recovery of China volumes is not an AI-related risk but is the primary near-term financial swing factor.

The Keytruda biosimilar risk is the company-defining challenge. When biosimilars enter — initially in Europe closer to 2028, then in the U.S. depending on patent litigation outcomes — the revenue erosion will follow Humira's trajectory but at 2.5 times the scale. The race to replace Keytruda revenue with next-generation oncology assets is the central question, and AI is accelerating competitors' ability to develop those assets.

Cost Exposure

Merck's R&D investment reached approximately $14.2 billion in 2024, representing 23% of revenue — among the highest ratios in large-cap pharma, reflecting the urgency of pipeline-building to offset the Keytruda cliff. AI efficiency in this context has a dual mandate: reduce the cost per successful program and increase the number of programs that can be prosecuted simultaneously.

The company's internal AI platform focuses on three areas: (1) target and biomarker identification using large-scale genomic and proteomic datasets; (2) molecular design for ADCs, where the linker chemistry and payload optimization are highly amenable to computational design; (3) clinical trial optimization through adaptive design and real-world patient matching.

ADC development is Merck's most AI-intensive pipeline activity. Its partnership with Daiichi Sankyo (three ADC programs, upfront and milestone payments totaling up to $22 billion) was itself facilitated by AI-enhanced due diligence on the ADC payloads and tumor expression data. AI tools for ADC linker stability, payload potency, and target antigen expression mapping are reducing the cost of ADC candidate generation by an estimated 40–50% versus historical biological assay-only approaches.

Moat Test

Keytruda's moat is clinical data scale, not mechanism novelty. The drug is approved in over 40 tumor types with hundreds of clinical trials supporting each indication. This data moat means that oncologists have extraordinarily granular outcome data for pembrolizumab across disease stages — data that biosimilar manufacturers cannot replicate and must rely on for substitution.

However, next-generation checkpoint combinations (PD-1 plus LAG-3, as in Opdualag from Bristol Myers Squibb) are generating superior response rates in specific cancers. AI-assisted patient stratification is allowing competitors to identify the precise tumor microenvironments where combination therapy outperforms PD-1 monotherapy. As these combinations become standard of care, Keytruda's monotherapy indication share in those tumors declines — not because biosimilars displace it, but because combination regimens replace it as the frontline recommendation.

Timeline Scenarios

1-3 Years (Near Term)

Keytruda continues growing into new indications — adjuvant and neoadjuvant settings in breast cancer, bladder, and early-stage lung cancer represent meaningful volume expansion even as individual indication competition grows. Winrevair establishes a PAH franchise with $1–2 billion annual revenue. Gardasil China volumes partially recover. AI-enhanced ADC programs (MK-2870 plus DXd payloads) advance to Phase II with initial efficacy signals. R&D efficiency savings of $400–600 million annually from AI tool deployment help hold R&D below 23% of revenue despite absolute spending increases.

3-7 Years (Medium Term)

Biosimilar pembrolizumab enters European markets (2028). U.S. entry timeline depends on patent litigation, likely 2028–2030. Merck must demonstrate that its own next-generation IO combination, whether a bispecific or an IO plus ADC regimen, produces superior outcomes to biosimilar pembrolizumab in key indications. This is the critical commercial bridge. AI-assisted ADC development from Daiichi Sankyo programs (HER3-DXd, B7-H3-DXd) must generate Phase III approvals to replace $5–8 billion in annual revenue.

7+ Years (Long Term)

Merck's long-term survival as a top-5 pharma company requires successful multi-wave pipeline execution. The company will face the Keytruda cliff with either: (a) a dominant next-generation IO platform built on AI-enhanced combination therapy design, or (b) aggressive M&A to acquire pre-built franchises. AI accelerates both pathways — making internal development faster and increasing the number of acquisition targets with de-risked pipelines.

Bull Case

Keytruda expands into early-stage settings (adjuvant/neoadjuvant) across 6–8 new tumor types before biosimilar entry, adding $4–6 billion in peak annual revenue. Daiichi Sankyo ADC programs generate 2–3 FDA approvals by 2029, establishing a $6–8 billion oncology franchise. Winrevair achieves $3–4 billion in PAH and potential HFpEF indications. AI-accelerated discovery yields a novel IO combination (PD-1 bispecific) that enters Phase III by 2027 with first mover advantage in next-generation immuno-oncology.

Bear Case

U.S. Keytruda biosimilar entry in 2028 triggers 30–40% price erosion within 24 months. ADC combination programs face unexpected safety signals in Phase III, delaying key launches. Gardasil China recovery stalls permanently below $3 billion annually. R&D spending must increase to 26–28% of revenue to fund the pipeline response, compressing operating margins from 37% to 28–30%. The company enters a negative operating leverage cycle as revenue declines faster than cost reductions.

Verdict: AI Margin Pressure Score 7/10

Merck scores 7 out of 10 — significant AI margin pressure driven primarily by the acceleration of oncology competition that threatens Keytruda's replacement therapy timeline. The Keytruda cliff is the largest single-asset revenue risk in large-cap pharma, and AI is compressing the window in which Merck must build replacement franchises. The company has the financial resources and pipeline assets to navigate this transition, but the margin trajectory through 2030–2032 will be genuinely challenging.

Takeaways for Investors

Merck is a high-stakes turnaround story disguised as a blue-chip compounder. The next three years of Keytruda growth are relatively predictable; the 2028–2032 transition is the value question. Investors should monitor: (1) Keytruda new indication approvals, particularly adjuvant settings that extend the volume runway before biosimilar entry; (2) ADC Phase III enrollment rates and interim data from MK-2870 and HER3-DXd programs; (3) Winrevair prescription trends as a leading indicator of its cardiovascular franchise potential; (4) any AI-accelerated pipeline readouts from Merck's bispecific antibody programs; (5) M&A activity, which accelerates meaningfully as the Keytruda cliff approaches. The stock's historically strong performance may not survive a 2028–2030 period of simultaneous biosimilar entry and pipeline execution risk.