AI BIOTECH × COMMERCIAL VALIDATION

AI Drug Discovery Is Making Money, but Drug Success Remains Unproven

Future World Signal | Issue 032 | 22 September 2026

Insilico Medicine has shown that AI-assisted drug assets can generate real revenue before approval. One profitable period, however, does not prove repeatable partnerships, Phase III success, or durable royalties and free cash flow.

Core company: Insilico Medicine (HKEX: 3696)

PUBLIC RESEARCH | FOUNDATION PHASE

CORE VIEW

The 10-second view

Insilico Medicine’s first-half profit shows that an AI-assisted discovery platform and the drug assets it creates can generate real revenue through licensing and research partnerships. This is business-model validation, not product-approval validation. The unanswered questions are whether deal revenue can repeat, Phase III can succeed, and eventual sales can create durable royalties and free cash flow.

$106.3 million

First-half 2026 revenue, up 287.2% year over year.

$35.54 million

Net profit for the period, not proof of durable free cash flow.

About 97%

Revenue from discovery and pipeline development, not drug sales.

Browser audio | Tap to play

Read the audio transcript

Insilico Medicine reported first-half revenue of one hundred six point three million dollars and net profit of thirty-five point five four million dollars. Large pharmaceutical companies are now paying for AI-assisted drug assets and research efficiency.

Revenue composition matters more than the profit headline. Roughly ninety-seven percent came from drug discovery and pipeline development, while software remained a small contributor. Growth was driven primarily by upfront and milestone payments.

The Lilly collaboration has potential value of up to two point seven five billion dollars, but the confirmed upfront payment was one hundred fifteen million. The rest depends on research, regulatory and commercial outcomes, so the deal ceiling is not revenue.

AI can prioritise targets and generate and optimise molecules, shortening early discovery. Human trials still determine safety and efficacy. The lead pipeline has reached Phase Three but remains unapproved.

Our view is that business-model validation has emerged, while final clinical and long-term value validation remain incomplete. The next evidence is repeat partnerships, Phase Three results, approval, royalties and durable free cash flow.

RESEARCH BRIEF

The 1-minute research brief

Insilico reported first-half 2026 revenue of $106.3 million, a 90.3% gross margin, net profit of $35.54 million and roughly $79 million of operating cash inflow. Cash and cash-management reserves stood at approximately $584.8 million.

Revenue composition matters more than the profit headline. Drug discovery and pipeline development generated about $103.1 million, or roughly 97% of revenue. Software solutions contributed about $2.7 million. The company said growth was driven primarily by upfront payments from new partnerships and milestones from existing ones.

The Lilly collaboration carries potential value of up to $2.75 billion, but the confirmed upfront payment was $115 million. The remainder depends on research, regulatory and commercial milestones, with royalties tied to eventual sales. A deal ceiling is optionality, not recognized revenue.

Our view: AI drug discovery has moved from “Can the technology be sold?” to “Can partnerships repeat, can clinical results replicate, and can milestones become royalties and sustainable free cash flow?”

01 · FINANCIAL SNAPSHOT

The profit is real, but revenue quality needs unpacking

Revenue$106.3m

Gross margin90.3%

Net profit$35.54m

Adjusted profit$51.23m

Operating cash inflowAbout $79m

Cash and cash-management reserves$584.8m

Revenue mix

Drug discovery and pipeline developmentAbout 97%

$103.1m

Software solutionsAbout 2.5%

$2.7m

The small difference between segment figures and total revenue reflects disclosure presentation and rounding. Software-solutions revenue should not automatically be treated as recurring subscription revenue.

02 · DEAL ECONOMICS

$2.75 billion is a ceiling, not cash already received

Confirmed upfront$115m

Higher-certainty consideration at signing

Potential deal valueUp to $2.75bn

Most of the value is conditional

Future royaltiesBeyond the ceiling logic

Require approval and actual sales

A contract ceiling measures possibility; confirmed payments and durable cash flow measure realised value.

03 · INDUSTRY PROGRESS

Which layer of AI drug discovery is actually validated?

Technology

Demonstrated

AI supports target prioritisation and molecule generation and optimisation.

Engineering

Strengthening

Candidate nomination typically takes about 12–18 months.

Clinical translation

Partial

The lead pipeline has reached Phase III but remains unapproved.

Commercial

Clearly emerging

Large pharmaceutical partners are paying upfront and milestone consideration.

Long-term value

Unproven

Approval, sales, royalties and durable free cash flow remain ahead.

04 · EXPLANATION AND VIEW

How can a company profit before its drug is approved?

Pharmaceutical companies do not need to wait for a product launch before paying. They can acquire development and commercialisation rights, faster target and molecule discovery, and the option to keep funding a programme in stages.

A typical agreement can include an upfront payment, development milestones, regulatory milestones, commercial milestones and sales royalties. Later payments can be larger, but their conditions are harder. A profitable period therefore validates the commercial value of assets and a platform; it does not mean the drug has been approved.

05 · EXPLANATION AND VIEW

Why is a $2.75 billion deal not $2.75 billion of revenue?

The Lilly agreement is a global licensing and research collaboration. Up to $2.75 billion represents the ceiling across potential payments, while the confirmed upfront consideration was $115 million. The remainder depends on research, regulatory and commercial outcomes.

Investors should not assign a 100% probability to conditional milestones or extrapolate one large upfront payment into permanent annual profit. A better framework separates upfront cash, stage-specific probabilities, future royalties and the research spending required to reach each gate.

06 · EXPLANATION AND VIEW

Which part of drug development does AI actually accelerate?

The company says its platform typically takes about 12–18 months to nominate a preclinical candidate. That advantage applies to early discovery; it does not mean a medicine can complete clinical development or obtain approval in a year.

AI can help prioritise targets and generate and optimise molecules, while automated laboratories can speed experimental feedback. Human safety, efficacy, long-term side effects and regulatory review still require staged evidence. AI can shorten the search for a drug, but it cannot erase clinical failure risk.

07 · EXPLANATION AND VIEW

What should investors actually watch?

One agreement shows that one buyer will pay for an opportunity. Expanded work with existing partners and sustained additions of new partners are stronger evidence of repeatable platform value. One profitable period matters, but revenue concentration, milestone conversion and cash generation across deal cycles matter more.

Repeat partnerships: Do existing partners expand and do new partners keep arriving?
Clinical replication: Do Phase III endpoints, safety and sample size support earlier signals?
Revenue quality: Do milestones, software renewals and royalties grow beyond upfront payments?
Cash retention: Can operating cash fund clinical development and become free cash flow?
Failure-aware valuation: Is conditional deal value probability-weighted?

08 · EXPLANATION AND VIEW

Who may benefit—and who may fall behind?

Potential beneficiaries include platforms combining proprietary data, models, automated laboratories and diversified clinical pipelines, together with large pharmaceutical companies using external licensing to improve research productivity. Laboratory automation, clinical-trial and biological-data infrastructure may also benefit.

Likely laggards include companies with impressive model demonstrations but no experimental or clinical assets, cash-constrained single-asset biotechs, and businesses whose long-term profit narrative depends on one upfront payment.

09 · NEXT STAGE

Value must be earned one gate at a time

01Repeat deals

02Milestone delivery

03Phase III evidence

04Regulatory approval

05Product sales

06Royalties and free cash flow

10 · RISKS AND WATCHLIST

What could invalidate the view?

Key risks

Irregular upfront payments can make revenue and profit highly volatile.
Lead programmes can fail, be delayed or require more capital.
Revenue can remain concentrated among a few partners.
Valuation can assign excessive probability to headline deal ceilings.
Data, intellectual-property or drug-regulatory rules can change.

What to watch

Expansion by existing partners and a continuing flow of new customers.
Conversion of signed programmes into later research and regulatory milestones.
Phase III confirmation of efficacy and safety in a larger population.
A shift from a few upfront payments toward more diversified, repeatable revenue.
Approved products, sales royalties and durable free cash flow.

Key concepts

Drug candidate: A molecule selected for further development but not yet approved.
Upfront payment: Consideration paid near the start of a partnership with relatively high certainty.
Milestone payment: A payment made only after a development, regulatory or sales condition is met.
Licensing: Rights granted to develop or commercialise an asset in a defined market.
Royalty: A percentage of future product sales paid to the asset owner.
Pipeline: A portfolio of drug programmes across research and clinical stages.

Frequently asked questions

How can Insilico profit without an approved drug?

Pharma partners can pay upfront and milestone consideration for licensing, research collaboration and staged development rights.

Has the full $2.75 billion been received?

No. The confirmed upfront payment was $115 million; most of the headline value remains conditional.

Does the profit prove that AI-designed drugs are more likely to win approval?

No. AI may improve early discovery efficiency, but human trials still determine safety and efficacy.

Is this revenue recurring?

Not in the same way as subscription revenue. Repeat partnerships, milestone delivery and future royalties must still be demonstrated.

What would materially strengthen the thesis?

Successful Phase III evidence, repeat partner revenue and a transition toward approved products, royalties and durable free cash flow.

What is the most common investor mistake?

Treating maximum deal value as received revenue, or extrapolating one large upfront payment into every future year.

Sources and research scope

Published on 22 September 2026. Financial figures are based on the 2026 interim results; clinical status follows public company materials and trial registration. Maximum contract value is not recognised revenue, and one profitable period is not durable earning power.