When AI Changes A Deal: Rethinking Risk, Milestones, And Timing In Life Sciences M&A

By Sally Wagner Partin, Sharon R. Flanagan and Torrey Cope

Artificial intelligence is increasingly embedded in drug discovery and clinical development, and early data suggests it is beginning to change how quickly companies reach early clinical milestones and how likely they are to clear them. For life sciences M&A practitioners and corporate development teams, the critical question is what these early signals mean for valuation, diligence, milestone design, and timing of transactions.

This article examines how AI is beginning to reshape drug development — particularly at the stages of target and compound identification — and explores the implications of those changes.

AI Is Already Changing Early-Stage Drug Development And Target Identification

The most immediate impact of AI is occurring at the earliest stages of drug development. AI-enabled approaches are improving target identification, molecule design, and lead optimization, allowing companies to move from concept to candidate more quickly. In practical terms, AI reduces the time and cost required to reach a defined candidate and enter early clinical development. By contrast, its impact on later-stage development — particularly on clinical efficacy outcomes — has been more limited to date.

Recent data reflects these divergent early results. While McKinsey reports reductions in development costs of up to 50% and accelerations of more than 12 months in clinical timelines, these benefits have not yet translated into improved success rates beyond Phase I. In a limited study of AI-native companies (i.e., companies that employed AI-driven methods from initial target identification rather than applying AI only at later clinical stages), Phase I success rates reached 80–90% for AI-enabled assets, compared to historical averages of 40–65%, while Phase II outcomes remained broadly consistent with historical norms at approximately 40%, according to a Boston Consulting Group analysis. PitchBook has similarly projected an overall increase in success rates, while indicating that gains have been concentrated in earlier stages of development.

Early indications are that weaker drug candidates — particularly those with early safety or developability issues — are being filtered out earlier, increasing the likelihood that programs entering the clinic clear initial safety hurdles. If more programs are able to reach and clear Phase I, that milestone may become less differentiating when comparing assets as a signal of underlying quality, even as it remains an important step in development.

AI Shows Promise For Additional Impact On Later Stages, And FDA Has Set the Stage For This Reality

As AI technology itself continues to advance, the ability to make early predictions of Phase II or Phase III success could improve soon. At a minimum, AI also has promise to accelerate, or at least reduce costs of, later-stage development. For example, AI may be used on early-stage data to identify specific types of patients that may be more likely to experience serious side effects (and therefore reduce the need for monitoring), or more likely to experience better outcomes (and therefore reduce the size of a trial needed to demonstrate efficacy).

FDA has expressed strong support for such uses of AI to expedite drug development and has recently put that into action by fleshing out a framework for determining when the technology can be relied on to provide evidence of drug safety, effectiveness, or quality. Although the demonstrated impact of AI has to date been limited primarily to Phase I success rates, this means that further impact is likely to become a reality.

Combined, these changes have direct implications for how assets are evaluated, when parties transact, and how transactions are structured and priced.

Seven Key Considerations For Dealmakers

1. AI is currently redistributing, not eliminating, development risk.

AI-driven improvements at the front end of development increase the likelihood that programs reach and clear early clinical milestones. By contrast, Phase II and later stages — where clinical efficacy and commercial value are determined — become relatively more important. Clinical risk is not eliminated; rather, it may be deferred or redistributed. As a result, early-stage milestones may become less differentiating across assets, and later-stage validation may become more central to value.

2. AI provides sellers increased optionality on deal timing.

AI-driven efficiency gains create competing effects on transaction timing rather than a single directional shift. Early-stage companies often sell because they lack the capital or resources to advance programs independently. If AI reduces the cost and time required to reach key clinical milestones, companies may be able to extend their cash runway by advancing fewer, more promising programs further before pursuing a sale. AI can also reduce the number of months a seller must operate at its burn rate before reaching a value-inflection point, and in certain parts of the discovery and preclinical process, may reduce that burn by enabling smaller teams and less reliance on external vendors. Together, these effects can extend runway without requiring additional capital. At the same time, faster early progress may allow companies to demonstrate credibility sooner, bringing assets to licensors or buyers earlier in their lifecycle.

The result is increased optionality for sellers. Some companies may choose to advance programs further before engaging in a sale process, while others may bring assets to buyers more quickly once they are able to demonstrate credible progress. What is already changing is the cost, speed, and volume of assets reaching the stages at which transactions typically occur. And there is potential for further reduction in the overall cost of development, particularly for the more expensive later stages of development.

3. Legal due diligence expands from clinical risk to data and model considerations.

Where a drug candidate has been identified or optimized using AI, the core diligence questions extend beyond whether the asset shows promise in the clinic to how the asset was generated. One key consideration is the likelihood that FDA will find the output of AI credible and therefore supportive of regulatory decision-making. Assessing this question should become more tractable as FDA’s framework for evaluating AI credibility takes effect and developers gain visibility into how it is being applied in practice.

Another key consideration is that defects in how a target or compound was identified can affect a buyer’s ability to use, defend, or commercialize the asset. Buyers will need to understand how the underlying data used to identify the asset was obtained and whether it can be lawfully relied upon, including whether patient data or clinical datasets were used within the scope of applicable consents and whether contractual restrictions could limit downstream use. This also raises questions about ownership and use of data and outputs generated in the discovery process, particularly where third-party data, tools, or collaborations were involved. In practice, this is likely to require more intensive front-end diligence and more detailed representations and warranties.

4. Early-stage milestones, pricing, and value may shift as Phase I becomes more predictable.

For early-stage life sciences acquisitions, milestone payments are often tied to Phase I initiation, frequently on an indication-by-indication basis. That framework may persist, but its economic significance may change.

Reaching Phase I has historically served as a meaningful indicator that an asset has cleared significant scientific and technical hurdles. That significance has not diminished; rather, as AI increases the probability that assets reach and clear Phase I, the value associated with that milestone may increasingly be priced in earlier and reflected in upfront consideration rather than contingent milestone payments. Buyers may be less willing to ascribe substantial incremental value to first-in-human dosing as a stand-alone event, while sellers — having demonstrated that early-stage risk has been meaningfully reduced — may seek a correspondingly greater share of deal value at signing.

Milestone achievement triggers may also evolve, shifting from initiation to completion or successful completion of Phase I, preserving their relevance while reflecting higher baseline probabilities of reaching and entering the clinic.

At the same time, if early-stage development becomes less capital-intensive and more programs are able to reach Phase I at lower cost, the number of Phase I-ready assets may increase. That dynamic may place downward pressure on pricing at that stage, or at least change how buyers and sellers assess value. Buyers may leverage increased competition from a larger pool of assets, while sellers — having deployed less capital and time — may be able to accept lower absolute prices while still generating attractive returns.

Similarly, to the extent AI reduces the cost, burden, and risk of later-stage development, sellers may argue that milestone structures should reflect that reduced investment by allocating a greater share of value to later-stage payments, rather than discounting those payments to account for development costs borne by the buyer.

As a result, value may concentrate more heavily at Phase II and later stages, where clinical efficacy — and therefore commercial viability — is first meaningfully established. Of course, if AI’s impact meaningfully improves Phase II success rates and beyond, where value concentrates may continue to move.

5. Milestone timing may compress.

AI may affect the timing of milestone payments even where milestone structures are unchanged. If earlier-stage milestones are reached more quickly, the time between signing and achievement may shorten. While AI does not yet appear to materially shorten the duration of clinical trials themselves, it may accelerate the path to those trials and transitions between them. As a result, assumptions regarding milestone timing reflected in valuation models and contractual sunset provisions may require reassessment.

6. Commercially reasonable efforts may evolve as AI becomes embedded in development decision-making.

Commercially reasonable efforts obligations may also evolve as AI becomes more embedded in development decision-making. Depending on the formulation, CRE is evaluated either by reference to what a similarly situated company would do (an objective standard) or by reference to the efforts a party applies to its own comparable assets (a subjective standard). AI has the potential to affect both frameworks.

To the extent AI reduces the cost and time required to advance early-stage programs, the baseline for what constitutes “reasonable” conduct may shift, as it may become more difficult to justify delaying or foregoing incremental development steps that can be undertaken more quickly or at lower cost. At the same time, if AI increases the number of viable programs competing for internal resources, parties may have greater flexibility — particularly under a subjective standard — to reallocate efforts toward higher-priority assets, and to discontinue or deprioritize programs earlier in development based on faster or more data-driven assessments.

The increasing availability of AI-enabled approaches may also influence how efforts are evaluated. If these approaches become widely adopted and are shown to improve development decision-making, a question may arise as to whether a party’s failure to use them is consistent with commercially reasonable efforts. At the same time, reliance on newer or less-validated methods may itself be challenged, particularly where those methods lead to decisions (such as deprioritizing or discontinuing a program) that would have been assessed differently using more traditional non-AI approaches. These dynamics may also place greater emphasis on how development decisions are made and documented, including the data inputs, analytical steps, and outputs generated in the course of using AI, and may make it easier to compare how a party has treated similar assets across its portfolio, including whether comparable programs were advanced or deprioritized under similar circumstances. To the extent those processes are recorded and reproducible, parties may create a more concrete record against which to evaluate whether commercially reasonable efforts were in fact applied. Because AI’s impact is currently more pronounced in earlier stages of development, these shifts may also be more relevant at early stages than later ones, where traditional cost and risk considerations continue to play a larger role.

7. Phase I and other early-stage failures may attract increased scrutiny and disputes.

If AI increases the probability that programs reach and successfully complete Phase I and early stages of development, failures at that stage may become less common but more notable. Where a program fails to meet what may be perceived as a higher baseline probability of success, that outcome may attract increased scrutiny from counterparties and increase the likelihood of dispute.

Specifically, missed early-stage milestones — which are frequently the subject of post-closing disputes — may be more likely to be contested where achievement was viewed as highly probable. In such cases, parties may more closely examine and litigate whether contractual standards, including commercially reasonable efforts obligations, were satisfied. At the same time, where a failure is clearly attributable to the underlying science rather than development effort, the stronger baseline filtering associated with AI may make commercially reasonable efforts claims harder to sustain.

Why This Matters For Dealmakers

At its core, life sciences M&A is an exercise in allocating development risk over time. AI does not eliminate development risk, but it is changing where that risk sits in the development timeline. For M&A practitioners and corporate development teams, that shift matters because milestone structures are designed to allocate and price uncertainty. If early milestones become more likely to be achieved and less economically differentiating, while later milestones continue to determine value, the balance of where and when consideration is paid may begin to change. For example, as early-stage outcomes become more predictable but less differentiating, transaction structures may increasingly shift value from early milestones to later-stage validation.

Understanding that shift, and reassessing whether early milestones continue to reflect meaningful risk or are instead priced into upfront consideration, will become an increasingly important part of how life sciences transactions are negotiated. This includes evaluating whether milestone triggers and associated timelines remain appropriate in light of potentially faster early-stage development; how commercially reasonable efforts obligations should be interpreted where development processes are increasingly data-driven; and whether diligence and representations adequately address the role of data and AI-enabled approaches in the identification and development of assets. Ultimately, as AI reshapes how programs are generated, advanced, and evaluated, practitioners who incorporate these considerations into their approach to valuation, diligence, and contract drafting will be well positioned to structure transactions that reflect — and appropriately allocate — the evolving dynamics of modern drug development.

About The Authors:

Sally Wagner Partin is a partner in Sidley's M&A practice who advises leading technology and life sciences companies on complex public and private transactions, corporate governance, and strategic business matters. She has represented clients in many of the industry's most significant acquisitions and is widely recognized for her work in corporate M&A.

Sharon Flanagan is a member of Sidley's Management and Executive Committees, managing partner of the firm's San Francisco office, and a leading M&A lawyer with more than 30 years of experience advising life sciences and technology companies on complex mergers and acquisitions, capital markets transactions, and corporate governance matters.

Torrey Cope is a partner in Sidley's Global Life Sciences practice who advises pharmaceutical, biotechnology, and medical technology companies on complex FDA regulatory issues across the product lifecycle, from product development through commercialization. He also serves as co-leader of Sidley's Life Sciences Transactions initiative, counseling clients on strategic collaborations, financing, licensing agreements, and M&A transactions.