Is Novo Nordisk Teaching Away Its Future?

By David A. Williams

Novo Nordisk’s agreement with Copenhagen-based Orbis Medicines is being presented as a conventional pharmaceutical collaboration: a large company pays a smaller, more specialised company to discover medicines that it cannot develop as quickly by itself. The potential value is considerable. Orbis could receive up to $1.4 billion in upfront and milestone payments, as well as royalties on future sales, while Novo will make a strategic investment in the company. Yet the most important part of the announcement is not the headline figure. It is Orbis’s declaration that its artificial intelligence platform “continuously learns” from the experimental data generated during drug discovery. That raises a question investors should be asking now rather than after the first successful medicine emerges: is Novo merely buying drug candidates, or could it also be helping to train an intellectual-property engine that it will never fully own?

The distinction matters because AI changes the economics of pharmaceutical collaboration. In an ordinary licensing agreement, the object being exchanged is relatively identifiable. One company develops a molecule, another purchases the right to develop or commercialise it, and the contract determines who receives what if the medicine succeeds. An AI-enabled discovery partnership is more complicated. Every compound designed, every failed experiment, every unexpected chemical interaction and every decision made by Novo’s scientists may become useful information. Unsuccessful molecules are not simply discarded. Their failures can teach the system which chemical territory to avoid, while successful experiments guide it towards more promising candidates. The collaboration may therefore do more than produce medicines. It may improve the machine that produces medicines.

At the centre of the agreement are macrocycles, relatively large, ring-shaped molecules that occupy the territory between conventional small-molecule pills and much larger biological medicines. Their unusual structure can allow them to reach disease-causing proteins that ordinary tablets struggle to affect. The difficulty has always been making them stable, effective and capable of being absorbed when swallowed. If Orbis can overcome those barriers, some conditions that currently require injections could eventually be treated with tablets.

Orbis says its proprietary nGen platform combines generative AI, automated chemistry, high-speed screening and a rapidly expanding experimental dataset. It begins with digital libraries containing as many as 100 billion possible compounds and can physically synthesise and analyse up to 100,000 distinct macrocycles within weeks. The platform does not merely ask a computer to imagine promising medicines. It makes real compounds, tests them in the laboratory and feeds the results back into the system. Each round of testing helps it make better choices during the next one.

That is precisely why the ownership question cannot be treated as a legal footnote. Novo possesses something Orbis does not: more than a century of accumulated knowledge about diabetes, obesity and cardiometabolic disease. It has clinical data, biological insight, manufacturing experience, regulatory knowledge and an understanding of why apparently promising molecules so often fail inside real patients. If even a carefully limited portion of that expertise enters the collaboration, it may help Orbis refine both individual drug candidates and the wider capabilities of nGen. Novo could receive valuable licences to medicines arising from the selected targets while Orbis retains the platform-level learning that made them possible. Novo might own some of the fish, but not the increasingly intelligent net.

There is currently no public evidence that Novo is surrendering its patents, proprietary clinical data or core intellectual property. The announcement does not disclose which company will own newly generated compounds, improvements to the platform, project data, AI refinements or discoveries falling outside the agreed targets. It would therefore be unjustified to claim that Novo is giving its intellectual property away. But the absence of disclosed terms does not make the question unreasonable. It makes the question unavoidable. In an AI discovery agreement, the commercially decisive clauses may concern not only patents and licensing rights, but also who can reuse experimental results, whether project data may improve the general platform, and whether knowledge acquired from Novo’s programmes may later benefit work conducted for another pharmaceutical company.

This is particularly sensitive because Orbis is not an ordinary external supplier. The company was founded by Novo Holdings, the investment company that is also the controlling shareholder of Novo Nordisk, and Novo Holdings remains an investor. Orbis’s subsequent €90 million financing round brought in other investors, including Lilly Ventures, the corporate venture-capital arm of Novo’s greatest commercial rival, Eli Lilly. Lilly Ventures’ investment does not in itself give Eli Lilly access to Novo’s research, Orbis’s confidential data or the results of this collaboration. Its presence does, however, illustrate how closely connected the competitive and investment networks surrounding pharmaceutical AI have become.

There is nothing inherently improper about this network. Indeed, it demonstrates the strength of Denmark’s life-sciences ecosystem. Capital, scientists and ideas circulate between established pharmaceutical companies and ambitious biotechnology ventures. But this circulation also makes the boundaries between Danish strategic capability, shareholder value and corporate ownership more difficult to see. A technology can be Danish in origin, supported by Danish capital and developed with Danish scientific expertise without necessarily being controlled by the pharmaceutical company contributing some of the most valuable knowledge to its development.

Novo’s direct investment in Orbis may help protect its position. If the platform becomes more valuable because of the collaboration, Novo will participate in some of that financial growth as a shareholder. Yet an equity investment is not automatically the same as control over intellectual property. The size of Novo’s proposed holding and the rights attached to it have not been disclosed. Unless the investment carries substantial governance, exclusivity or contractual protections, owning shares in Orbis will not necessarily give Novo control over the platform, its accumulated data or its future applications.

The alternative argument is that Novo is not teaching away its future at all. It may instead be purchasing time, which has become one of the pharmaceutical industry’s most expensive commodities. Drug development is slow, failure is common and Eli Lilly has demonstrated how quickly competitive advantage can move in the obesity market. Novo cannot reasonably insist on inventing every technology internally while more agile biotechnology companies explore new areas of chemistry. Paying for access to an existing platform may be far more rational than spending years trying to reproduce it.

If Orbis can turn treatments resembling injectable biologics into effective tablets, Novo gains access to something with profound commercial potential: medicines that may be easier for patients to take, simpler for healthcare systems to administer and less dependent on injection devices. Oral treatment does not automatically mean cheap or easy manufacturing. Sophisticated macrocycles may still be technically demanding and expensive to produce. They could nevertheless alter the distribution and administration of chronic treatment, particularly if they reduce dependence on sterile filling, injection training, specialist delivery equipment or cold-chain logistics.

The company that succeeds in replacing selected injections with oral macrocycles could expand treatment to much larger populations. Novo is therefore not merely buying another experimental obesity drug. It is purchasing a position in a possible transition from traditional biological treatments to computationally designed chemistry. The commercial prize is not simply a better pill. It is an entirely new way of attacking biological targets that conventional pills have been unable to reach.

The danger is that Novo may value the visible products more highly than the invisible learning system behind them. Pharmaceutical companies are accustomed to measuring pipelines by the number of candidates entering clinical trials. AI platforms accumulate value differently. Their most important asset may be the growing relationship between molecular design, experimental failure and biological outcome. Once knowledge from thousands of experiments has been absorbed into the system, its selection methods and its proprietary datasets, it becomes difficult to distinguish what the platform knew before a collaboration from what it learned during one. Traditional contractual language may struggle to capture an intelligence whose most valuable output is not a single invention, but an improved ability to invent.

Novo should therefore be able to answer several questions, even if commercially sensitive details remain confidential. Will data produced through the partnership be used to improve Orbis’s general platform? Who owns improvements derived from experiments funded by Novo? Can those improvements be applied to programmes involving other pharmaceutical companies? Does Novo receive exclusive rights only to specified molecules and targets, or also to related chemical families and later discoveries? If Novo supplies biological data or specialist knowledge, can the lessons extracted from that information influence work beyond the collaboration? And if the platform becomes indispensable, does Novo possess sufficient investment rights, licensing protections or acquisition options to avoid becoming permanently dependent on technology developed outside its direct control?

These questions should not be mistaken for hostility towards collaboration. They represent the necessary scrutiny of a new pharmaceutical economy in which discovery itself is becoming a learning process. The old fear was that a company might license away a successful molecule too cheaply. The new fear is subtler: a company may contribute the experiments, clinical understanding and scientific judgement that make another organisation’s discovery platform smarter, then find itself paying repeatedly for access to intelligence it helped create.

The Novo-Orbis agreement could become a model of Danish industrial cooperation. It joins Novo’s cardiometabolic expertise with a home-grown biotechnology company using Danish scientific talent, automated laboratories and AI-supported molecular design. If it succeeds, patients could gain oral treatments for diseases still dominated by injections, Orbis could become an internationally significant drug-discovery company, and Novo could rebuild its pipeline around a new class of medicines. That is the optimistic reading, and it is credible.

But there is another possibility. Novo may be placing some of its most valuable scientific problems inside an external system whose underlying intelligence will remain proprietary to Orbis. The medicines produced for Novo may be protected, while the lessons extracted from developing them strengthen a platform capable of outliving any individual licensing agreement. The decisive question, then, is not whether Novo has given away its intellectual property in the conventional sense. There is no public evidence that it has. The question is whether the company has adequately protected its share of the new knowledge created when its scientific experience, biological understanding and research priorities are used to educate somebody else’s machine.

In the age of AI-designed medicine, ownership of the drug may no longer be enough. The deeper prize is ownership, or at least meaningful control, of the system that learns how to discover the next one.

Sources

Orbis Medicines: Novo partnership announcement and financial terms

Orbis Medicines: nGen platform and Copenhagen expansion

Orbis Medicines: platform capabilities, financing and management

Fierce Biotech: analysis of the Novo-Orbis agreement