Pharma's nine-figure data+AI deals: why not just write a service contract?

The deals are about human architecture just as much as the tech.

2026 has produced another run of them. Noetik, Immunai (1, 2, 3), Chai Discovery (1, 2, 3, 4). These are companies whose product is some combination of proprietary biology data and the models trained on it. Where figures are available, the upfronts run into the tens of millions, with headline values and milestones reaching into the hundreds of millions and beyond.

Now look at what's actually being purchased.

A defined set of targets. A research term of three to five years. Access to a dataset and the models trained on it, usually non-exclusive. Deliverables that a data-and-model provider in any other industry would happily scope, price, and invoice against.

Which is to say: a large fraction of this could be written as a service contract. Statement of work, defined outputs, defined term, a few million dollars a year at most. Pharma has procurement functions that write exactly this kind of contract every week, and they are very good at it.

So why doesn't pharma just do that?

My take

The premium isn't buying better science. It's buying mutual commitment to a result.

And that commitment is the specific mechanism by which new technology actually gets adopted inside a large pharma organization as opposed to being evaluated, admired, and sidelined.

Consider the technology pilot

A team runs it. It has a champion. It has a budget that fits under someone's discretionary line, which is the entire reason it got approved in six weeks instead of nine months.

It produces a readout. The readout is usually fine: promising, directionally positive, worth a follow-on.

Then the champion moves to another therapeutic area, or takes a role at a competitor, or gets a new set of priorities. The pilot stops being anyone's job. The vendor sends a check-in email and gets a warm reply about the overall excitement and revisiting it next fiscal year.

A service contract, though it does carry a greater commitment and financial weight, ends the same way. It renews or it doesn't. And though there may be department head visibility on it, it rarely goes up past the VP level.

Now consider the partnership

A deal with staged milestones carries much more weight. It has a joint steering committee, which means named people on both sides with recurring obligations. It has a governance cadence that outlives any individual champion. It has a research plan and a budget that appears in a board deck.

Most importantly, it has milestones. This means both parties have agreed, in writing, on what counts as something coming out of this. The partner is committed to delivering against them. The pharma is committed to being in a position to evaluate them, which means allocating internal scientists, data infrastructure, and decision-making attention that a service contract would never command.

And it produces a second-order effect that matters more to this audience than the deal terms do. Someone senior on the data side has to sit on that committee and defend the work - to a therapeutic area head, to finance, to the partner across the table. That's a structural promotion of the data function, and it happens at signature, independent of whether the science eventually works.

What's actually being bought

None of these are drug assets but Pharma buys them anyway, at prices well above what the same work would cost as a service, because it isn't buying the work. It's buying a structure in which the work has to be defended, reviewed, and acted on by people whose names are on it.

The human architecture around the deal: the committee, the cadence, the signed plan, the budget line, is what converts a capability into something that gets implemented. 

Dig deeper into the topic 

At DataDrivenPharma West 2026 Day 2 with a panel on “Data as an Asset in Pharma and Biotech Deals” with Eka Korthonkia (J&J), Matthias Thurner (Genentech), Michelle Chen (Form Bio), and Pinar Bailey (Fenwick & West).

Check out our website for #ddpwest26 passes, and for discounts on 3+ non-commercial and 2+ commercial teams.

Also, this week only, get $75 off with the code 75OFF (ends Aug 14 end of day)

See you at #ddpwest26!

Best,

Ilya Captain

DataDrivenPharma Founder