Who takes the risk of innovation when the pharmaceutical industry doesn't?

In the first half of 2026, pharmaceutical and artificial intelligence companies announced fourteen major deals totaling at least US$23.6 billion in potential value, according to a survey by analyst Oliver Sprague Kelly. The prevailing interpretation was that the pharmaceutical industry is betting heavily on radical innovation .

I spent a good part of my twenty-five years in the industry in environments where such agreements are discussed and approved, and those who have lived through this learn to read the announcement from back to front, starting with the upfront payment. In the largest of these, between Lilly and Insilico Medicine , the announced value was up to US$2.75 billion and the upfront payment was US$115 million, or 4.2%. This is not an exception: in AI drug discovery partnerships, upfront payments have averaged around 2% of the headline figures, a ratio of about 50 to 1 between what was promised and what was actually transferred.

This isn't bad faith; it's a sound fiduciary design in the face of high technological risk, and I myself have defended such structures. But it means that the pharmaceutical industry isn't taking the risk of radical innovation. It's buying, cheaply, the right to take advantage of it if it works.

The question that matters, then, is who took over? And the answer, which is hardly ever discussed here, is that it was the state .

The money that made Insilico exist.

In August 2022, Insilico Medicine closed a Series D2 funding round led by Prosperity7 Ventures , the growth fund of Aramco Ventures , the investment arm of the Saudi state oil company, bringing its total Series D funding to US$95 million . The capital was earmarked for advancing the pipeline, building a fully automated discovery lab, a biological data factory, and establishing regional centers. The company's main program was, at that time, in Phase 1.

It's worth noting the dates. State-funded capital financed Insilico's operation in 2022, when the company was in its early stages, without any approved assets and without any guarantee that its platform would work. Lilly arrived almost four years later, in March 2026, paying 4.2% upfront for an option on still pre-clinical assets. Between those two moments, someone else bore the technological risk. It wasn't the pharmaceutical company .

And the link to national strategy is explicit, not incidental. Under Vision 2030 , the Saudi government has a declared focus on developing its technology sector, with approximately US$20 billion allocated solely to artificial intelligence. The quid pro quo was also explicit: Insilico expanded its presence in Saudi Arabia and established regional bases in the Middle East. There was no philanthropy involved. There was industrial policy, equity participation, and anchoring of capacity within its own territory.

This is not an isolated case; it's established doctrine.

When you look closely, the pattern appears everywhere.

Isomorphic Labs , a spin-out of Google DeepMind , raised US$2.1 billion in May 2026 in a round led by Thrive Capital , with Alphabet and GV , and with three states at the same table: MGX , from Abu Dhabi , Temasek , from Singapore , and the UK Sovereign AI Fund , from the United Kingdom . Here too, the money comes in before the results are achieved, to finance the company's entry into the clinical market, and not to reap a return that already exists.

MGX , created in 2024 by Mubadala and G42 and chaired by the UAE's national security advisor, closed its first fund at US$49 billion , with positions in OpenAI, Anthropic, and xAI, and aims to surpass US$100 billion under management. Saudi Arabia is building domestic capacity through HUMAIN, Qatar is prioritizing infrastructure, and Singapore is buying direct stakes in border laboratories. A recent survey on sovereign wealth funds accurately summarizes the movement: they are being used by governments to execute national strategies and build stronger positions in global value chains, and the study mapped twelve new vehicles, including Ireland , Great Britain , Botswana , and Spain .

The British design deserves particular attention because it is the most transferable to countries that don't have tens of billions available. The UK Sovereign AI Fund invests in equity like a venture capital fund, with an explicit mandate to generate returns for the taxpayer. But the check is the smallest part of the offer. Along with it come up to 1 million GPU hours on national supercomputers, expedited visas to attract international talent, access to curated national databases, and access to public contracts.

This isn't just funding. It's an operational package that simultaneously addresses capital, computing power, talent, and first customer acquisition, precisely the four areas where a healthcare deep tech company typically fails. And it's a package that only the state can assemble because three of the four components aren't available on the market.

The trade-off is real, on both sides.

It's important to state clearly that none of this is state generosity, nor is it an unreturnable public risk.

A state that enters early buys a stake in a company that, if it goes public, is worth many times more. It buys territorial anchoring , as Saudi Arabia did by linking its entry to the expansion of Insilico in the country. It buys a position in a value chain that will shape the health of the next two decades. And, in the British case, it operates under a declared mandate of financial return , not under a logic of granting nothing in return.

From the company's perspective, the reward is equally concrete. The relationship between Lilly and Insilico began in 2023 with a software licensing agreement, progressed to a $100 million partnership in November 2025, and reached $2.75 billion in March 2026, in addition to the IPO in Hong Kong. Sanofi made a similar move with Earendil Labs , with an agreement in April 2025, another in January 2026, and participation in the company's own funding round. Renewal and scaling are worth much more than size, because the first is earned through presentation and the second through results.

What this architecture organizes, therefore, is not a heroic startup single-handedly bearing the burden of radical innovation. It is a division of labor in which the State finances the phase that no one else finances , remains in the capital, and reaps the rewards when the industry finally appears to buy the option. The risk is public from the start, and the return is shared at the finish line .

The risk of innovation: The necessary conversation

It would be unfair to say that Brazil hasn't moved forward. I've already written here about the valley of death for innovation in Brazilian healthcare and how the instruments have multiplied in recent years, including economic subsidies, support for initial clinical research, a private equity fund dedicated to healthcare, and state programs that have corrected the logic of technological maturity-based pathways. There is new money and a better design than before.

But all these instruments belong to the same family. They are ways of financing projects . What Saudi Arabia , the Emirates , Singapore , and the United Kingdom have built is of a different nature: they are ways of taking a position and paying part of the bill with assets that the State already owns and that the market does not sell.

This distinction seems to me the most useful point for the Brazilian debate at this moment. A grant finances a stage. An equity stake follows a trajectory . A package like the British one solves, in one fell swoop, the missing capital, the computing power the company can't afford, the talent it can't attract, and the first contract it can't win on its own.

It's worth noting that Brazil does indeed possess assets of this nature. We have public computing capacity, we have large-scale public databases, and above all, we have the largest public health system in the world, whose purchasing power is rarely treated as an instrument of innovation, and is almost never offered to a startup as a potential first client. These are assets that already exist, have already been paid for, and remain, to a large extent, off the table.

We are experiencing a rare moment of public attention to radical innovation in healthcare in Brazil, with a newly established national program and important decisions still pending on how it will be implemented. It is precisely at times like these that it is worthwhile to look at what countries that have decided to take this seriously have actually done, and not just what they have announced.

So I return to the question in the title. When the pharmaceutical industry doesn't take the risk of radical innovation, and it doesn't, the State does, in places where someone has decided it's worthwhile. Not out of generosity, but because they understood that whoever finances the journey has the right to be at the finish line. The question that remains is whether we are willing to have this conversation here, while there are still decisions to be made.

At the Brazilian Institute for Innovation in Health , we closely monitor precisely this boundary: the discrepancy between what advertisements promise and what contracts deliver, and between who bears the risk of innovation and who decides whether to share it. If this type of analysis resonates with your work or your institution, we would be happy to discuss it.

by Marcio de Paula

Founder of the Brazilian Health Innovation Institute - IBIS, with over 25 years of experience in life sciences, he has held strategy and innovation positions in companies such as Biolab and Ferring Pharmaceuticals , founded the Brazilian Pharmaceutical Innovation Network and is a member of health innovation boards in Brazil. He writes about the paths to radical innovation in health and how to connect science, industry, and public policy.