The pharmaceutical industry's reliance on external innovation has reshaped drug development. Emerging biopharma companies (those with under USD 200 million in annual R&D spend, including pre-commercial firms and those with up to USD 500 million in annual revenue) originated 41 of the 48 novel active substances (NAS) launched in the United States in 2024, and 59% of NAS over the 2020–2024 period.1IQVIA Institute, Global Trends in R&D 2025 (IQVIA, March 2025) 8–12. University spin-offs, research hospitals, and biotechnology startups drive the therapeutic pipeline. Yet the contractual frameworks governing these collaborations often lag behind the scientific ambition they enable.
1. Who Bears the Downside When Collaborations Fail?
When a multinational pharmaceutical company partners with an academic institution or emerging biotech, the power imbalance is structural. The pharma partner brings capital, regulatory expertise, and market access. The academic partner contributes scientific innovation and often the foundational intellectual property. But who bears the consequences when Phase III trials fail, or when they succeed in unexpected ways?
Standard collaboration agreements address milestone payments, publication rights, and commercialization terms. What they frequently underestimate is the allocation of sunk costs when projects terminate unexpectedly or pivot substantially. Agreements that lack express provisions for sunk-cost recovery tend to default to the rules on unjust enrichment (Art. 62 ff OR), which may produce outcomes neither party intended. And when a collaboration produces valuable platform technology that neither party anticipated at the outset, the absence of prospective IP allocation clauses can freeze development while ownership is litigated.
Under Swiss law, the characterization of such collaborations carries concrete consequences. If a court treats the arrangement as an einfache Gesellschaft under Art. 530 ff OR (a partnership in which both parties pursue a common purpose), fiduciary duties attach to both sides, losses are shared equally absent contrary agreement (Art. 533(1) OR), and each partner acquires joint ownership (Gesamteigentum) of contributions made to the joint venture. A pure license-service structure, by contrast, limits each party's liability to its own contractual obligations: the licensor delivers rights, the licensee pays fees, and neither bears the other's losses. Innominate or hybrid arrangements fall somewhere between, with courts assessing the economic substance of the relationship rather than its contractual label.
The distinction matters most at the point of failure. When a collaboration structured as a license terminates, the licensee walks away from the licensed technology. When the same collaboration is recharacterized as a simple partnership, both parties may face claims for shared losses, and the academic partner may discover that intellectual property contributed to the joint purpose has become partnership property subject to liquidation under Art. 548 ff OR. Recharacterization also reaches the clauses the parties relied on to limit their liability: a negotiated limitation-of-liability clause allocates risk between the contracting parties, but the equal loss-sharing rule of Art. 533(1) OR applies between partners of a simple partnership independently of any contractual cap, so the cap may not survive the recharacterization. Agreements that do not provide for recharacterization can leave smaller partners with losses their business models cannot absorb.
Many collaboration agreements also leave intellectual property ownership unresolved. Under Art. 332(1) OR, inventions made by employees in the course of their work and in performance of their contractual obligations belong to the employer (Art. 332(2) OR provides only a right of acquisition for inventions outside contractual duties), but in academic collaborations, the "employer" may be a university whose IP policy reserves certain rights to the institution or the individual researcher.2Art. 332 OR (SR 220); cf ETH Zurich, Regulations on Intellectual Property (RSETHZ 440.4, 2005). When the academic partner is a US institution, the Bayh-Dole Act (35 USC §§ 200–212) lets universities elect to retain title to federally funded inventions while preserving the government's march-in rights. Under those rights, the government may require licensing to third parties on four statutory grounds. One is a failure to take effective steps to achieve practical application of the invention, a term the statute defines to require that the invention's benefits be available to the public on reasonable terms.335 USC §§ 200–212; 35 USC § 201(f); 35 USC § 203; NIST, Draft Interagency Guidance Framework (2023) 88 Fed Reg 85593. A Swiss biotech that licenses foundational IP from a US university must therefore contend with the possibility that the US government could intervene in the licensing chain, a risk that has no analogue under Swiss employer-invention rules and that standard collaboration agreements rarely address.
The most commercially valuable output of a failed clinical program may be the data explaining why it failed, yet most collaboration agreements leave that data unallocated.
2. When Milestone Payments Create Unintended Obligations
The pharmaceutical industry has developed sophisticated milestone-based payment structures. These mechanisms assume a relatively linear progression through development phases. Modern drug development increasingly involves adaptive trial designs, basket trials, and platform approaches that do not fit neatly into traditional milestone frameworks.4US Food and Drug Administration, Adaptive Designs for Clinical Trials of Drugs and Biologics (FDA, December 2019).
The mismatch is structural: agreements that define milestones by regulatory phase rather than value creation events tend to misallocate risk in adaptive-design programs. A platform technology generating multiple candidates simultaneously may trigger no milestone at all under a phase-gate structure, even as it creates substantial commercial value. Similarly, when regulatory pathways shift mid-development (through breakthrough therapy designations, accelerated approval, or rolling submissions), a rigid phase-based framework either overpays for truncated development or leaves the innovating partner uncompensated for accelerated value delivery.
The EU Clinical Trials Regulation5Regulation (EU) No 536/2014 on clinical trials on medicinal products [2014] OJ L158/1. and FDA's adaptive trial guidance contemplate development pathways that may compress, expand, or run in parallel.
Swiss-based biotechs collaborating with EU or US pharma partners rely on Swiss freedom of contract but may also meet mandatory provisions in their partners' jurisdictions. A milestone structure that is enforceable under Swiss law may encounter public policy limitations when performance occurs in jurisdictions with different regulatory philosophies.
3. Who Owns the Data When Clinical Programs Fail?
Perhaps the most underappreciated aspect of R&D partnerships concerns the treatment of negative results. Failed compounds generate valuable data about mechanisms of action, patient populations, biomarkers, and safety signals. Many collaboration agreements focus exclusively on successful outcomes, leaving ownership and utilization rights for negative data undefined.
Who owns the learnings from a failed trial that could inform future research directions, and does ownership transfer with program termination or remain with the party that generated the data?
Whether an academic partner can publish findings from a terminated program without commercial partner consent depends on how publication rights were structured, yet many agreements address publication only for successful outcomes. The tension between confidentiality obligations and the scientific community's interest in negative results remains largely unaddressed in standard collaboration templates.
And when safety signals emerge from failed programs, what obligations exist to share that data with regulatory authorities across jurisdictions, obligations that may override contractual confidentiality provisions entirely?
Under the revised Swiss Bundesgesetz über den Datenschutz (DSG), in force since September 2023, the processing of clinical trial data that involves personal information is subject to specific requirements that may conflict with broad data-sharing provisions in collaboration agreements.6Bundesgesetz über den Datenschutz (DSG) of 25 September 2020 (SR 235.1), in force 1 September 2023. Contractual data rights, regulatory reporting obligations, and data protection requirements can pull against each other, and few agreements reconcile all three.
For collaborations involving EU clinical trial sites, data transfer complexity depends on the ultimate destination. Switzerland benefits from an EU adequacy decision, so transfers from EU trial sites to Swiss sponsors do not require Standard Contractual Clauses (SCCs) or supplementary measures.7Commission Decision 2000/518/EC [2000] OJ L215/1; Case C-311/18 Schrems II; EDPB, Recommendations 01/2020 (2021). Onward transfers are different: when Swiss sponsors share clinical data with US-based affiliates or CROs that are not certified under the Swiss-US Data Privacy Framework (DPF; for certified recipients, Swiss law has recognized an adequate level of protection since 15 September 2024), those transfers trigger the Schrems II supplementary measures analysis, and collaboration agreements must anticipate them.
4. Why Governing Law Choices Are Rarely Neutral
The choice of governing law in pharma R&D agreements is rarely neutral. Swiss law, New York law, and English law represent the three dominant choices, each carrying distinct implications for contract interpretation, remedies, and dispute resolution.
Swiss law offers relative flexibility in contract formation and interpretation, with courts generally respecting party autonomy.8Art. 19(1) OR (n 2) (Vertragsfreiheit); Art. 18–19 IPRG (SR 291). But two distinct mechanisms under the IPRG constrain that autonomy in multi-jurisdictional collaborations, and the distinction between them matters. Art. 18 IPRG requires Swiss courts to apply Swiss overriding mandatory provisions (lois d'application immédiate) regardless of the chosen governing law, provisions such as competition law restrictions or, potentially, data protection requirements under the DSG. Art. 19 IPRG goes further: it permits Swiss courts to take into account overriding mandatory provisions of a foreign law where the situation has a close connection with that jurisdiction, a discretionary power with a high threshold, but one directly relevant when clinical trial activities occur in EU Member States whose regulatory frameworks impose their own mandatory requirements. For multi-jurisdictional clinical programs, this means that choosing Swiss governing law does not insulate the agreement from mandatory rules of the jurisdictions where trials are actually conducted.
Dispute resolution clauses in pharma collaborations frequently specify arbitration (ICC, Swiss Rules, or ad hoc arrangements). Arbitration offers confidentiality advantages in an industry where commercial relationships and regulatory standing are intertwined. But arbitration's limited discovery mechanisms may disadvantage the party with less access to information, typically the smaller innovator. And emergency arbitrator procedures, while faster than court proceedings, may not be fast enough when development timelines are at stake.
What happens when a collaboration agreement specifies Swiss law and ICC arbitration, but the dispute concerns conduct at a US clinical trial site that may also trigger FDA enforcement concerns? Standard arbitration clauses are not drafted for a dispute that a regulator may take up on the same facts.
5. What Happens When a Partner Walks Away Mid-Trial?
Termination provisions in pharma R&D agreements deserve particular scrutiny. Standard "termination for convenience" clauses may seem flexible, but their exercise can be existential for smaller partners. An emerging biotech that has hired a clinical team, leased laboratory space, and committed two years of runway to a single collaboration may find that a termination-for-convenience notice, exercised because the pharma partner's portfolio priorities shifted, leaves it with sunk costs it cannot recover and a pipeline gap it cannot fill before its next funding round closes.
Notice periods that do not track the development phase can produce disproportionate outcomes: a 90-day notice that is adequate during preclinical work may be grossly insufficient during a Phase III trial where patient enrollment, site contracts, and regulatory submissions are in flight. Transition assistance provisions (covering data transfer, regulatory filing continuity, and ongoing safety reporting) are essential but frequently omitted or drafted in generic terms that provide no operational guidance. And valuation of partially completed work remains among the most contentious termination issues: the party closer to the next milestone has every incentive to overstate proximity, while the terminating party has equal incentive to discount it.
Under Swiss law, an open-ended Dauerschuldverhältnis (continuing obligation) may be terminated by ordinary notice of reasonable length absent contrary agreement and, where continued performance has become unreasonable for one party, ended early for good cause (aus wichtigem Grund).9On ordinary termination with reasonable notice, by analogy Art. 546 OR (n 2) (six months' notice for a simple partnership of indefinite duration); on extraordinary termination for good cause, BGE 138 III 304 E. 7; BGE 128 III 428 E. 3c. In the context of a clinical development program, what counts as "reasonable" notice (and what rises to "good cause") must account for patient safety wind-down, regulatory notification requirements, and the smaller partner's ability to secure alternative funding or partnerships, factors that the partnership provisions applied by analogy were not developed to address. Whether the collaboration agreement creates a simple partnership (with its associated fiduciary duties and dissolution procedures under Art. 545 ff OR) or merely a set of bilateral obligations determines both the exit mechanisms available and the standard of conduct required during the wind-down period.
For agreements governed by Swiss law but involving EU regulatory submissions, termination may trigger marketing authorization transfer obligations under EMA procedures.10Regulation (EC) No 726/2004, Art. 16b; Directive 2001/83/EC [2001] OJ L311/67; European Medicines Agency, Transfer of marketing authorisation (EMA, 2024). A party may therefore hold the contractual right to exit and still face regulatory obstacles to exercising it.
6. What Is Changing, and What It Means for Agreement Design
Two developments are changing the conditions under which these agreements operate. First, the trend toward consortium-based R&D models (where multiple academic institutions, biotechs, and pharma companies collaborate on shared platforms) multiplies the contractual complexity geometrically.11IHI (Innovative Health Initiative), EU-EFPIA public-private partnership (EUR 2.4 billion budget, 2021–2027); EFPIA, The Pharmaceutical Industry in Figures (2025). Bilateral frameworks become inadequate when three or more parties contribute IP, share data, and co-own regulatory submissions. Consortium agreements require governance structures, voting mechanisms, and deadlock-resolution procedures that bilateral templates simply do not contemplate.
Second, the revision of the EU pharmaceutical legislation, initiated by the Commission's April 2023 proposal to replace Directive 2001/83/EC and Regulation (EC) No 726/2004, will alter the regulatory incentive structures around which collaboration agreements are built. The European Parliament adopted its first-reading position in April 2024 and the Council agreed its negotiating position on 4 June 2025, but with trilogue negotiations still to reconcile the two texts, key parameters remained open. Changes to data exclusivity periods, orphan drug protections, and environmental risk assessment requirements will affect milestone definitions, commercialization timelines, and the economic models underpinning partnership terms. Agreements drafted during this legislative cycle must build in sufficient flexibility to accommodate a framework whose final requirements remained undefined as of September 2025.12COM(2023) 192 final and COM(2023) 193 final (26 April 2023).
Against this background, the issues examined in this article (contractual characterization, milestone design, IP ownership, data rights, governing law constraints, and termination provisions) depend on one another. A milestone structure that appears sound in isolation may produce perverse incentives when combined with termination-for-convenience rights and ambiguous IP reversion clauses. A governing law choice that seems neutral may expose the entire agreement to mandatory provisions neither party anticipated. Each provision may be sound on its own while the agreement as a whole does not account for how they act together. That shows most at the point of failure, where the gaps between template assumptions and operational reality decide the outcome.
These are not issues that standard templates resolve. They require analysis calibrated to specific partnership structures, therapeutic areas, and jurisdictional configurations, analysis that begins where template drafting ends.