A US biotech acquires a Basel research operation, or licenses a compound discovered there, and its counsel runs the intellectual-property diligence the way it always has. The representations recite that the target owns its inventions. The employment agreements contain an assignment clause. The chain of title reads as closed. That confidence rests on a default US counsel rarely states out loud, because in the United States it barely needs stating: an employee's inventions become the employer's once the employee has assigned them, and competent drafting makes the assignment automatic. Swiss law does not begin in the same place.
1. The Default US Counsel Imports
The US baseline is more constructed than it feels from inside it. An invention initially belongs to the natural person who made it; the employer secures title through an invention-assignment agreement, and patents themselves are personal property assignable by written instrument.135 U.S.C. § 261: patents are personal property, assignable by written instrument. For federally funded research, the Bayh-Dole framework adds a narrower route, letting a nonprofit organization or small business firm elect to retain title to inventions made with federal support, though it reaches only that federal-funding slice rather than the ordinary private R&D relationship.2Bayh-Dole Act, 35 U.S.C. §§ 200–212; § 202 on federally funded inventions. None of these routes is self-executing, and how title actually moves has been worked out case by case. The Supreme Court has held that Bayh-Dole does not itself vest title in the contractor; it presupposes that the inventor's rights were first assigned. The common law supplies a narrower default of its own, under which an employee hired to invent who succeeds at the very task must assign the resulting patent even without an express clause. And where the parties rely on the clause, the Federal Circuit has made the outcome turn on its precise words: a present assignment, in which the employee "hereby assigns" inventions not yet made, moves title automatically as each invention comes into being, while a clause that merely has the employee "agree to assign" in the future leaves the employer with an equitable expectancy rather than legal title.3Stanford v Roche (2011); Standard Parts v Peck (1924); FilmTec (Fed Cir 1991).
The operative point is not that US assignment is fragile, but that it is a drafting achievement rather than a statutory gift. US counsel who have internalized this still carry a workable mental model into a cross-border transaction: the inventor owns first, the paper cures it, and a clean assignment clause is the instrument that closes the gap. That model travels intact into deals involving research performed abroad, where it is applied to Swiss-generated inventions as though the only variable were the quality of the drafting. The Swiss position is not a stricter version of the same rule. It is a different allocation, and the difference is structural rather than one of degree.
2. How Art. 332 OR Sorts Employee Inventions
Swiss law, like US law, begins with the inventor: the entitlement to a patent belongs to the inventor, to the successor in title, or to a third party to whom the invention belongs on another legal ground, the last being the doorway employment law uses.4Patentgesetz (PatG) vom 25. Juni 1954 (SR 232.14), Art. 3. What differs is the mechanism that moves an employee's invention to the employer. In Switzerland that mechanism is supplied by employment law, not by the assignment agreement, and it does not treat all employee inventions alike. Art. 332 OR (the Swiss Code of Obligations) produces three categories, two of them expressly and the third by what it leaves untouched, and only one of the three behaves the way the US model predicts.5Obligationenrecht (SR 220), Art. 332 (Rechte an Erfindungen und Designs).
Service inventions, those an employee makes, or contributes to making, both in the course of official activity and in fulfillment of contractual duties, belong to the employer by operation of law, whether or not they are protectable. No assignment is needed and no separate payment is due. For a scientist engaged to invent, this category usually captures the output, and it reaches non-patentable inventions alongside patentable ones, and designs as well as inventions, so the allocation is not confined to the patent estate.
Contingent inventions are the ones that unsettle the US assumption. These are inventions an employee makes in the course of official activity but not in fulfillment of contractual duties, and they do not pass to the employer automatically. The employer can acquire them only if it reserved that right in writing, and the reservation is not itself the transfer: the employee has to give the employer written notice of the invention, and the employer then has six months to state in writing whether it wishes to acquire it or releases it to the employee. Where the invention is not released, a special, appropriate compensation falls due, whose amount turns on facts a deal model does not surface. Free inventions, those made outside the employee's official activity altogether, form the third category: they remain the employee's, and the employer has no acquisition right in them at all.
A US invention-assignment clause promises that everything the employee invents already belongs to the employer. Swiss law answers that some of it does, some of it does not, and some of it must be paid for.
The compensation attaching to contingent inventions is not a default term the parties can quietly rewrite. It counts among the relatively mandatory provisions of Swiss employment law, so a clause purporting to exclude or reduce it to the employee's disadvantage is ineffective to that extent.6Art. 362 OR: the Art. 332(4) OR compensation cannot be waived to the employee's disadvantage. The dividing line between service and contingent inventions appears, at first, to resolve cleanly for a research employee: someone engaged to invent is fulfilling contractual duties, and the invention falls on the automatic side. The difficulty is that the category is fixed by the employee's actual duties, not by the language of the assignment clause. An invention by a process engineer, a data scientist recruited for analytics rather than discovery, or a clinical-operations lead who solves a formulation problem can fall on the contingent side, where automatic vesting never happened, however comprehensive the assignment paragraph appears. Roles also shift over time, and an invention made after a data scientist has drifted into discovery work may sit on a different side of the line than a contract drafted years earlier would suggest.
Art. 332 OR also has boundaries the US model does not prepare counsel to look for. It governs employees, and only employees: inventions made by an independent contractor, a consultant, or a contract research organization fall outside it entirely, governed by whatever the services contract says and by nothing the statute supplies, so a Swiss consultancy agreement that omits an assignment can leave title with the contractor rather than with the company that paid for the work. Whether a given individual is an employee or a contractor is itself a contested question under Swiss law, and a nominal consultant can be recharacterized as an employee, moving the same invention from one regime into the other. Whether Art. 332 OR reaches a given invention at all is not answered by the site of any single experiment either: the law governing the employment relationship follows the employee's habitual place of work, so a seconded or remote researcher, and an invention made on a trip abroad, do not line up neatly with a plain "Swiss-based staff" shorthand.
3. Where the Standard US Clause Misfires
A US present-assignment clause is engineered to move title the US way: the inventor owns first, the clause assigns, and the gap closes. Dropped into a Swiss employment relationship, it meets three categories and does something different in each. For service inventions it is redundant as to the Swiss vesting, which has already happened by operation of law, though it is not pointless: that vesting does not by itself close the chain of title in jurisdictions that lack an equivalent rule. For contingent inventions a broad written assignment can do real work, since a comprehensive clause can serve as the written reservation Art. 332 OR looks for; whether a clause drafted as an advance assignment also dispenses with the statutory notice and the six-month election is not something the provision settles, and what the clause certainly cannot do is extinguish the special compensation, which is not the employer's to waive to the employee's disadvantage. For genuinely free inventions a competently drafted US clause carves out part of the category, since several US states limit how far an assignment clause may reach into an employee's own time; the carve-outs are not the same shape, though, because those statutes commonly leave an own-time invention assignable where it relates to the employer's business or came out of work done for the employer, while Swiss law asks only whether the invention was made outside the employee's official activity. An invention a US clause validly captures can therefore be one Art. 332 OR leaves with the employee. Where the clause does reach beyond what Swiss law permits, the excess does not fail invisibly; the reach-through can be unenforceable to that extent, and the compensation entitlement can outlast the clause that tried to displace it.
Choice of law does not repair the drafting, and it has more layers than the contract's face suggests. An individual employment contract performed in Switzerland by an employee habitually working there is governed, absent a valid choice, by Swiss law, and the menu of laws the parties may even choose is itself narrow, confined to the law of the employee's habitual residence or of the state where the employer has its place of business, domicile, or habitual residence.7IPRG (SR 291): Art. 121 habitual-workplace rule and narrow choice menu; Art. 18 overriding mandatory provisions; Art. 110 protection-country rule; Art. 122(3) employer-employee IP agreements. For a Basel scientist employed by a Swiss entity, a Delaware or New York clause usually falls outside that menu, so Swiss law governs in full; and even a validly chosen foreign law does not sweep away Switzerland's overriding mandatory provisions, which apply regardless of the law the parties designated. Selecting Delaware or New York law therefore does not import the Delaware default; it leaves the Swiss protective core standing while creating the impression, on the face of the document, that it has been contracted around. A further layer sits beneath even this: whether the internal allocation of an invention also settles the proprietary position against each national patent, or whether that follows the law of the country for which the patent is granted, is genuinely debated, and can resolve differently from one patent to the next. The gap between what the paper appears to achieve and what it achieves is precisely the kind of gap that does not announce itself until someone tests it.
Nor is the Swiss position the German one, which US counsel sometimes treat as a stand-in for "the European approach." Germany's employee-invention statute requires compensation for service inventions themselves and runs a formal claiming mechanism that Swiss law does not use.8German ArbnErfG (1957): compensation for service inventions on a distinct claiming mechanism. A structure calibrated for German employee-invention law is no more transposable to Switzerland than a US one; "European" is not a single regime here, but at least three. Whether the reservation the statute wants exists in the specific contracts, whether the steps its acquisition depends on were ever completed, and whether any compensation was assessed or paid are questions that were rarely documented when the work was done, because the US-drafted paper made them look unnecessary to ask.
4. The Diligence Surface: Acquiring or Licensing Swiss IP
In a transaction, the ownership question is not asked invention by invention. It is compressed into a representation that the target owns its intellectual property, and that representation sits on top of Art. 332 OR for every invention generated by Swiss-based staff. For inventions that were genuinely part of a scientist's assigned work, the representation describes what the statute has already vested; the difficulty is telling which inventions those actually are, and the same ownership-of-research-output terrain is examined from a different angle in Insight 16. For the rest, the paperwork and the statute can describe different things, and what a deal inherits depends heavily on how the deal is built.
The starting point is the inventor-to-employer link, which matters whatever the structure: if a contingent invention never vested in the employer, because the contract was silent or narrowly drawn, the entity that appears to own it does not, and its patents can be exposed on standing as much as on title. How that defect then moves depends on the deal. A US biotech acquiring a Basel operation usually buys the shares of the Swiss entity, in which case nothing is assigned to the buyer at all: the IP stays inside the company, and any defect stays with it, inherited at the level of the equity rather than moving across a bill of sale. An asset purchase or a license is where a chain-of-title metaphor genuinely bites, because there the rights have to be traced and conveyed one by one, and a link that Swiss law never completed cannot be conveyed.
The compensation entitlement behaves differently again, and the distinction is easy to get wrong. It is the inventor's claim against the employing entity, an employment debt; it does not attach to the invention, so it does not travel with the IP itself to an asset-buyer or a licensee. In a share deal the liability is nonetheless acquired, not because it rides with the IP but because it sits inside the company whose shares changed hands; in a license it stays behind with the licensor. An asset deal is less settled than the transfer documents suggest, because Art. 333 OR carries an employment relationship, with all its rights and obligations, over to a buyer who takes the business or a part of it unless the employee refuses the transfer, and holds the previous employer and the acquirer jointly and severally liable for the employee's claims that fell due before the transfer and for those falling due until the employment could ordinarily have been terminated.9Art. 333 OR: on transfer of a business the employment relationship passes to the acquirer with all rights and obligations; Abs. 3 imposes joint and several liability. Which entity actually carries it is its own question in a group where a Swiss operating company employs the scientists while a separate holding company owns the patents, since the debt stays with the employer even after the IP has been assigned onward, so a buyer can acquire the patents in one entity and leave the liability in another. The exposure is also not open-ended: the claim is subject to a limitation period, though whether the shorter employment-claim period or the longer general one applies, and when it starts to run, are themselves contested,10Art. 127 / Art. 128 Ziff. 3 OR: the special compensation is subject to a limitation period; its length and start are contested. so the tail is bounded even if its length is not settled.
The licensing case has its own edge. When a US company in-licenses a compound or a platform developed in Switzerland, the operative question is whether the Swiss licensor owns what it purports to license. A licensor without clean title to a contingent invention grants rights it cannot fully stand behind, and an exclusive or sublicensable structure inherits the same defect at every tier beneath it. The compensation claim does not travel down that chain, since it runs against the employing entity rather than against anyone downstream of it in the license structure; what travels is the title weakness, which a license several transactions removed from the Basel laboratory can still sit on top of without curing.
A buyer might reason that the Swiss employees signed assignment agreements, so title is secure. That reasoning holds for the inventions the statute already vests, and is weakest precisely for the inventions it does not, which are also the inventions least likely to have been separately documented, because they arrived through work the employee was not specifically engaged to perform. It is complicated further by the way pharma patents are actually authored: a single patent commonly names several inventors, whose contributions can fall in different categories, and a defect in any one inventor's link, including a contractor or CRO co-inventor outside Art. 332 OR altogether, can cloud title to the whole. An indemnity can allocate the loss from a title defect between buyer and seller, but it cannot convert an invention the employee still owns into one the target owned, nor discharge a compensation entitlement the employee never waived. A position that looked closed at signing can reopen when the compound the invention underlies reaches value.
5. Strategic Considerations
The questions that matter here are not answered by confirming that assignment agreements exist. They begin one level below the paperwork, and they resist resolution in the abstract.
Which inventions were actually service inventions?
Because the category is set by the employee's real duties rather than by the assignment clause, whether a given invention vested automatically depends on how the role was defined and on what the employee was actually engaged to do, a characterization the standard clause does not settle and that a data room rarely records. Where a portfolio was built by a mix of discovery scientists and staff hired for adjacent functions, the automatic-vesting assumption may hold for some inventions and not for others, without any visible marker of which is which.
What did the employment contracts actually secure?
For contingent inventions, Swiss law looks for a written reservation, and while a broad written assignment can sometimes serve that role, a US-style clause that assumes the point is unnecessary, or that speaks only to inventions made in fulfillment of duties, may not reach the category at all. Whether the specific contracts do the work Art. 332 OR contemplates, rather than the work the template was drafted for, is a question about the actual paper in the specific matter, and it interacts with whether the compensation was ever addressed and the acquisition steps ever completed.
What compensation exposure has accumulated across the portfolio?
The compensation entitlement is the inventor's, cannot be waived to the employee's disadvantage, and can lie dormant, within its limitation period, until an invention becomes valuable. Whether it has been triggered, assessed, or discharged for any given invention, and how it aggregates across a research portfolio built over years, is the kind of latent liability that a purchase price allocation and a reps-and-warranties structure were not designed to surface on their own.
What did the choice-of-law clause actually achieve?
A US or other foreign governing-law clause in a Swiss employee's contract may achieve considerably less than its plain terms suggest. Whether it displaced the Swiss employee-protective core or merely papered over it, and what follows for the inventions generated under it, turns on the interaction between the contract, the place of work, and the mandatory provisions of Swiss employment law. These questions turn on the specific contracts, the specific duties, and the specific commercial history of the inventions at issue, and none of them is the kind of question a template answers.