• AI visibility

Guide · AI search

Can AI search help a biotech company find partners, licensees and investors?

It can help a biotech get onto more scouting lists, but only when its science, pipeline and partnering status are public, consistent and backed by sources an assistant trusts. Investors and researchers already use AI tools to find and summarize deals and studies, while pharma licensing hit a 10-year high in 2025. No study yet shows how often pharma dealmakers use public AI assistants to scout assets, so treat this as a reasoned bet, not a proven channel.

By Underneath Team Guide Updated

The short version

  1. Licensing is where much of a biotech’s money now comes from: J.P. Morgan’s Q4 2025 report, using DealForma data, counts $250.2 billion in announced biopharma licensing value across 516 deals in 2025, the highest since 2016, with upfront cash steady at 7% of deal value.
  2. Venture money is scarcer: IQVIA (opens in a new tab) puts 2025 US and European biotech venture funding at $24 billion across 410 rounds, down 14%, with only 11 US biotech IPOs.
  3. Buyers are hungry: over $230 billion of industry revenue faces loss of exclusivity by 2030, IQVIA estimates, and emerging companies hold 70% of clinical-stage assets, most of them unpartnered.
  4. The people who screen biotechs already lean on AI. In a PitchBook and Web Summit survey of 116 investors, the top uses were summarizing due diligence materials (34%) and identifying relevant deals (26%); in Elsevier’s survey (opens in a new tab) of 3,000 researchers, 58% use AI tools at work, up from 37% in 2024.
  5. When AI answers cite sources, independent ones dominate: in a study of ChatGPT’s health answers, Jacques and colleagues (opens in a new tab) found medical institutions supplied 30.6% of 615 cited sources and peer-reviewed journals 5.7%.

Who does a biotech actually need to reach, and what is a deal worth?

Pharma dealmakers, investors and scientific collaborators. One license can bring a nine-figure upfront payment.

A biotech has three audiences, and each controls a different kind of money:

  • Pharma business development and licensing teams. They decide whether to license an asset or buy the company. J.P. Morgan counts 41 licensing deals in 2025 with upfront payments above $100 million. IQVIA lists the largest: Pfizer paid a record $1.25 billion upfront for ex-China rights to 3SBio’s PD-1/VEGF bispecific antibody, and GSK paid $500 million upfront in a Hengrui alliance worth up to $12 billion.
  • Investors. Venture, crossover and later public investors fund the years before a deal. Money is tighter and more concentrated: Massachusetts biotech companies raised $6.85 billion of venture capital in 2025, down 13% and the smallest total since 2019, according to a MassBio report covered by Banker & Tradesman (opens in a new tab).
  • Research collaborators. Academic labs, hospitals and other biotechs bring data, patients for trials and validation. They also cite your work, which matters for the other two audiences.

Platform biotechs add a fourth: pharma companies that buy discovery services. The landmark example is AI itself; IQVIA notes that Lilly and NVIDIA committed up to $1 billion over five years to a joint AI drug discovery lab.

The demand side is strong. IQVIA estimates big pharma’s deal capacity at $1.3 trillion and calls the patent cliff a source of “mounting pressure” to replenish revenue. Competition for that money is global: 40% of assets big pharma in-licensed in 2025 had a Chinese licensor, IQVIA reports, and the value of deals for China-originated assets rose from $45 billion to $105 billion. A US or European biotech is no longer competing only with its neighbors for a buyer’s attention.

Where do AI assistants already show up in biotech dealmaking?

In investor screening and scientific literature research, documented in surveys; pharma scouting use of public assistants is unmeasured.

Investors. The PitchBook and Web Summit survey found 26% of investors use AI to identify relevant deals. S&P Global Market Intelligence’s 2025 outlook survey (opens in a new tab) of more than 100 private equity, venture and limited partner respondents found 31% saw due diligence as the most useful place for generative AI and 22% picked deal sourcing. Neither survey is limited to life sciences investors.

Scientists. Elsevier found that 61% of researchers using AI tools use them to find and summarize the latest research, and 51% use them for literature reviews. A collaborator looking for a delivery platform or a validated target may start with an AI summary of the field before reading papers.

Pharma dealmakers. Specialist scouting tools for business development teams are on the market, and data providers such as CB Insights (opens in a new tab) sell AI drug discovery scouting reports. We found no survey that measures how often licensing teams type questions into ChatGPT, Gemini or Perplexity. Our inference is that public assistants are used for early orientation (who works on a target, which modalities are crowded) while decisions still rest on paid databases, confidential data rooms and meetings.

That last step remains physical. BioSpectrum Asia’s preview of BIO 2025 (opens in a new tab) reported 20,000+ registrants, 10,000+ partnering delegates and 60,000+ partnering meetings expected in Boston. AI search does not replace those meetings. It can influence who asks for one.

What might a dealmaker or investor ask an AI assistant about your field?

Questions about targets, modalities, stages and who is still unpartnered. We wrote these examples ourselves; none were collected from real buyers.

AudienceIllustrative question
Pharma licensing team“Which biotechs have Phase 2 oral GLP-1 programs that are not yet partnered?”
Pharma search and evaluation“Who is developing PD-1/VEGF bispecific antibodies outside China?”
Venture investor“Which seed-stage radiopharmaceutical startups have published preclinical data?”
Crossover investor“What readouts are expected in in vivo CAR-T in the next 12 months?”
Academic collaborator“Which companies have lipid nanoparticle platforms that reach tissues beyond the liver?”
Platform customer“Which AI drug discovery companies have taken a molecule into the clinic?”

A single question rarely stays single. Google says its AI Overviews and AI Mode may use a “query fan-out” technique (opens in a new tab), running related searches on subtopics, and OpenAI says ChatGPT search rewrites a prompt (opens in a new tab) into one or more targeted queries. A question about a target can quietly become separate searches for trials, publications and recent deals. When our hidden-searches study logged ChatGPT’s searches, 43.8% of its answers included a search aimed at a named publication, ranking or award. In biotech, the named authorities are journals, trial registries and the trade press.

How does being named in an AI answer lead to a license or a funding round?

By adding your asset to a scouting list early, before the non-confidential deck and the partnering meeting request.

As we read the evidence, the path looks like this:

  1. An analyst, investor or scientist asks about a target, modality or indication.
  2. The answer names a handful of companies and cites the pages it used: publications, trial registry entries, press coverage, company pipeline pages.
  3. The reader checks the cited science and the company’s own pipeline page, then requests a non-confidential overview or a meeting at the next partnering event.
  4. Confidential diligence follows, then a term sheet: an upfront payment plus milestones for a license, a round for an investor, or a joint grant or data-sharing agreement for a collaborator.

The money sits at step 4, but the selection happens at steps 1 to 3. J.P. Morgan’s data show why the economics favor partnering now: licensing headline values rose mainly through milestone packages, while venture rounds concentrated in later-stage companies. For a preclinical or Phase 1 company, being found by a licensor may be the most realistic route to cash.

None of this shows up cleanly in analytics. A pharma scout who met your asset in an AI answer may contact you months later through a banker or a conference app. We discuss the attribution problem in how to prove GEO caused sales.

What makes an AI assistant cite one biotech and not another?

The companies do not publish their rules; studies point to independent, institutional and recent sources, which suits science.

Documented by the platforms. Google and OpenAI both say their AI answers search the web and show links to sources. Neither explains why one company working on a target is named and another is not.

Observed in studies.

  • In consumer health questions, ChatGPT drew most citations from institutions such as medical centers, government agencies and Wikipedia, Jacques and colleagues found; peer-reviewed journals supplied 5.7% of the 615 cited sources. That study covers patient questions, not dealmaking, but it shows how heavily assistants lean on institutional names in health.
  • For US software questions, AI search drew 72.7% of its sources from earned media, such as independent reviews and publications, Chen and colleagues (opens in a new tab) report. Biotech has no equivalent study.
  • In our study of brand entities, independent coverage was the strongest predictor we measured: each tenfold increase in the number of independent sites naming a brand in the cited pages went with 4.7 times the odds of being recommended.
  • Our freshness study found the four assistants cited pages first published about half as long ago as Google’s top 10 for the same questions (ratio 0.50). A pipeline page last updated two readouts ago works against you.

What scientists say they trust. In Elsevier’s survey, 74% of researchers called peer-reviewed research trustworthy, and 59% said an AI tool that automatically cites references would increase their confidence in using it. Biotech buyers think the same way, we infer: an AI answer that cites your paper or trial record is more useful to them than one that cites your homepage.

Our inference for biotech. The proof a licensing team checks is largely public or could be: publications, posters, trial registrations, patents, press releases with data, and coverage in trade outlets. An assistant can find and cite those same documents. A company whose science lives only in a confidential deck gives both the human scout and the assistant very little to work with.

What does a biotech lose when AI answers leave it out?

Mostly early meetings and inbound interest, though no study has priced that loss for biotech.

  • A crowded, global field. With 40% of big pharma in-licensed assets now coming from Chinese licensors, a scout comparing options for a target sees more candidates than before. An asset absent from the first summary must be found some other way.
  • Fewer funding routes. With US and European biotech venture funding at $24 billion and IPOs scarce, partnering is a funding source, not only an exit. Missing early scouting conversations can mean a longer, costlier runway, we infer.
  • Answers move. A single good answer is not a fixed position. Assistants disagree with each other and with themselves from run to run, as our guide to why AI answers about your brand change explains.
  • Wrong facts travel. An outdated stage, a discontinued program or a confused company name can be repeated in answers. Our guide to fixing wrong brand information in AI answers covers corrections.

What does GEO work look like for a biotech?

Making your science, pipeline and partnering status easy to find, cite and repeat accurately, without promises of placement.

  1. One consistent entity. Use the same company name, modality, targets, indications and stage on your website, registry entries, partnering-event profiles, investor materials and databases. Small companies with similar names are easy to confuse.
  2. A public, dated pipeline page. State each program’s target, modality, indication, stage and whether it is available for partnering, with the date of the last update and links to data.
  3. Citable science. Publish papers and posters, keep trial registrations current, and post data releases as plain web pages rather than only as slide PDFs. These are the sources researchers already say they trust.
  4. Independent coverage. Trade press articles, conference presentations and partner announcements give assistants third-party confirmation. The general method is in how brands build authority for AI search; notability for an encyclopedia entry is discussed in why Wikipedia matters for AI search.
  5. Plan around newness. Young companies and new programs are often missing from assistants’ answers, as we explain in why ChatGPT misses new products.
  6. Measure several assistants. Ask the questions a scout, investor or collaborator would ask in ChatGPT, Gemini, Perplexity, Claude, Copilot and Google’s AI features, repeatedly, and record which companies and sources come back.

Stay inside the rules. Overstated efficacy claims, selective data and planted coverage are scientific, legal and reputational risks for a company that will face diligence; see legitimate GEO versus manipulation. Many of the startup lessons in how AI startups win customers from AI search also apply to young biotechs.

Which parts of the AI scouting picture are still unproven?

The link between AI visibility and signed biotech deals is unmeasured; the surveys describe general AI use, not assistant-led scouting.

  • No pharma scouting data. We found no survey of how often business development teams use public assistants to find assets.
  • The investor surveys are broad. PitchBook’s sample is 116 investors attending one technology conference, and S&P’s covers private markets in general, not life sciences funds.
  • Health citation research covers patients. The ChatGPT health study looked at consumer questions; dealmaker questions may draw on different sources.
  • Deal totals differ by database. J.P. Morgan reports $250.2 billion of licensing value and IQVIA $232 billion for 2025, because they count deals differently. Both agree it was a decade high.
  • Nothing ties an AI mention to a term sheet. The wider evidence is reviewed in does AI visibility drive business results.

Where should a biotech start if it wants more partnering conversations?

Ask the questions your future licensees and investors would ask, see which companies are named, then fill the public gaps.

List the targets, modalities and indications you work on, and write the questions a pharma scout, a fund analyst and an academic collaborator would ask about each. Ask each question in several assistants, then ask it again on another day. Record which companies appear, which papers and registries are cited, and whether your stage and partnering status come back correctly. In biotech, the gap is usually science kept in a confidential deck instead of on citable public pages.

If you want help, talk to us about a partnering visibility review. It shows which of your programs surface when dealmakers and investors ask about your field, which sources the assistants rely on, and which missing public evidence is most likely costing you scouting calls, partnering meetings and term sheets. The generative engine optimization service page describes how we help move science out of the confidential deck and onto dated pipeline pages, citable data releases and independent coverage.

Frequently asked questions

Do pharma business development teams use ChatGPT to find assets?

No published survey measures it. Investors do use AI: 26% of investors in one survey use it to identify relevant deals.

Should a biotech publish partnering status on its website?

Usually yes, for programs you want to out-license. A clear, dated statement helps human scouts and gives assistants something accurate to repeat.

Do scientific publications help AI visibility?

Plausibly. Researchers trust peer-reviewed work, and AI health answers cite institutions and journals, but no study has tested biotech dealmaker questions.

Can a preclinical company be found by AI assistants?

It can, if its data are public and covered by independent sources. Very new companies are often missing, so expect months, not weeks.

Does GEO replace partnering events?

No. Events like BIO’s convention, with 60,000+ partnering meetings expected in 2025, remain where deals start. GEO can affect who requests a meeting.

Sources

Free strategy call

Some questions are easier to answer about your own business.

Bring the one that matters most. On a free 30-minute call we’ll take a first look at it and send you a short written read afterward.

Ask about my businessHow we work