TL;DR
- Probability of success (POS) is consistently overestimated in BD deals, distorting valuations before terms are signed.
- Most pharma partnerships fail; alliance management remains a critical and underweighted discipline.
- Big pharma is increasingly inclined to build internally rather than partner — unless someone else is faster.
- Valuation is shifting from comparable transactions toward eNPV, which puts probability modeling closer to the center of deal math.
- More rigorous, data-driven probability assessments could meaningfully reset how BD teams price and structure deals.
Most pharma partnerships fail. Probability of success is consistently overestimated. And big pharma is increasingly inclined to build internally rather than partner — unless someone else is moving faster. Those were a few of the recurring themes at the Pharma Partnering Summit, held April 29–30 in San Diego, where Evangelia Loizou, Ph.D., Senior Engagement Lead at Intelligencia, participated in a panel discussed with the topic: “Partnering between AI companies and pharma/biotech.”
She returned from San Diego with interesting insights we are sharing with you in this blog.
How are pharma and biotech actually de-risking deals?
The opening session zeroed in on reducing risk, with neurodegenerative disease singled out as particularly difficult to de-risk.
In this session, several recurring takeaways emerged:
- Small biotechs tend to do better in early stages, while bigger companies hold structural advantages in pipeline and governance. That makes handoff models attractive — a biotech might run IND-enabling work and hand the asset to a large pharma at Phase I.
- Breaking milestones into smaller pieces was flagged as a key de-risking tactic.
- The panelists flagged probability of success (POS) as a consistently overestimated metric across BD deals.
- Expected NPV (eNPV) was described as more important than comparable transactions when valuing a deal.
The POS-overestimation point is worth pausing on. If buyers and sellers are systematically too optimistic about an asset’s odds, that bias gets baked into deal terms before anyone signs. More accurate POS assessments could decrease this bias and ensure that deals are priced realistically.
What are partners actually looking for?
The afternoon panel, “What partners look for,” surfaced a more nuanced view of how big pharma evaluates opportunities.
Biology is still the foundation of any conversation, but the panel noted a clear distinction between “sexy” and “less sexy” modalities. Small molecules remain reliable, while biologics are beating standard of care in several indications.
A blunt comment from the J&J side captured a broader shift: rather than partnering, J&J prefers to build capabilities internally — unless someone else is doing something good and faster. The takeaway from the panel as a whole was that most partnerships fail, and that hearts, minds, and wallets all need to align for a deal to work.
The Biogen panel pushed further, emphasizing that success criteria for collaborations need to be defined explicitly, and that alliance management is a critical and often-underweighted discipline.
Which assets drew attention on the floor?
Three outstanding assets were highlighted in discussions:
- Phanes (San Diego) presented its bispecific antibody platform, with three Phase 2 assets in development, including a CD47 bispecific, a claudin program, and T-cell engagers. The company has dosed more than 300 patients and has secured three Orphan and three Fast Track designations.
- Sathgen (Princeton) presented a triple-negative breast cancer drug targeting an indication with roughly 40,000 new cases per year. The mechanism is CD44 stem-cell inhibition, and Phase 1a is complete. While an interesting approach, the data so far are not compelling.
- Protagonist discussed its psoriasis pill and signaled an in-licensing interest spanning immunology and inflammation, hematology, and metabolic disease. The company is focused on oral peptide drug development and targeting one to three high-quality programs.
Where does AI fit into the deal-making conversation?
Several of the themes connected directly to challenges associated with accurately assessing the probability of technical and regulatory success (PTRS) and the opportunities related to improving these assessments. A few stood out:
- POS overestimation. If buyers and sellers are systematically too optimistic, there is room for more rigorous probability modeling to recalibrate assumptions before deal terms are locked in.
- Milestone structures. The Sarepta and Arrowhead discussions both touched on milestone-based de-risking. Program-level probability data can inform how milestones are sized and sequenced. Scenario analysis adds another layer to that conversation.
- eNPV over comps. If the industry is shifting toward eNPV-driven valuation rather than reliance on comparable transactions, probability modeling becomes a more central input to deal math.
- In-licensing among mid-caps. Protagonist, VirA, Phanes, and Sathgen all surfaced in-licensing or partnering needs spanning peptides and HIV alongside oncology and immunology. Competitive landscape analyses in these therapeutic areas are therefore particularly relevant in gauging a compound’s potential in the market.
- Alliance management. The Biogen panel’s emphasis on understanding the partner’s actual need points to a broader role for shared evidence bases — including probability data — that both sides of a deal can work from.
What therapeutic areas are worth watching?
Several therapeutic areas surfaced repeatedly across the two days: neurodegenerative disease (with its de-risking challenges), triple-negative breast cancer, peptides, immunology and inflammation, metabolic disease, and hematology, were mentioned as TAs to watch.
The overall signal from San Diego: BD teams are sharpening their tools, partnerships are getting harder to land, and there is a need for better evidence on the questions that matter most such as the probability of success of a drug in development.
FAQs
What is PTRS?
PTRS — probability of technical and regulatory success — is the likelihood that a drug in development will progress from its current phase through to approval. It captures both the scientific risk that a compound fails in trials and the regulatory risk that it fails to gain approval. PTRS is typically expressed as a percentage and used as an input to valuation, milestone structuring, and pipeline prioritization.
How does PTRS support BD and deal-making?
PTRS gives both sides of a deal a quantitative basis for valuing an asset. It feeds directly into eNPV calculations, informs how milestones are sized and sequenced, and provides a shared frame of reference for negotiation. Because human judgment tends to overestimate POS, a more data-driven PTRS estimate can correct optimism bias before it gets baked into deal terms.
What does AI add to probability modeling?
AI-driven approaches can analyze large historical datasets — clinical trial outcomes, mechanism of action, indication, trial design, sponsor track record, and more — to generate probability estimates grounded in patterns across thousands of comparable programs. The result are estimates that are more reproducible, less subject to individual bias, and more useful as a shared evidence base across BD, R&D, and finance teams.
For more tailored, data-rich insights, let’s talk.
