What insurance does an AI drug discovery startup need?

AI drug discovery startups span two regulatory and underwriting paradigms simultaneously. The software/SaaS layer (the platform serving pharma customers) underwrites like enterprise SaaS — tech E&O, cyber, D&O, IP infringement. The biology layer (any wet lab work, biospecimen handling, or claims about therapeutic outcomes) underwrites like biotech — D&O sized to clinical-stage risk, professional liability, and depending on FDA classification, products liability.

The structural question for the insurance program is which layer dominates. A pure-play software platform serving pharma R&D customers underwrites primarily as enterprise SaaS with a biotech overlay. A platform that operates its own wet lab, generates biological data, or makes clinical claims underwrites primarily as biotech with a tech overlay.

Tech E&O for AI drug discovery startups serving pharma customers should be sized to the largest single customer contract liability cap. Enterprise pharma SaaS contracts frequently include liability caps at 1-2x annual subscription fees with carveouts for IP infringement, gross negligence, and breaches of confidentiality. Tech E&O limits typically run $2M-$10M for mid-market startups and scale up materially for late-stage operators with major pharma customers.

IP infringement is a material exposure for AI drug discovery startups — both algorithmic IP (claims that the AI/ML system infringes third-party patents) and output IP (claims that AI-generated molecular designs infringe drug discovery patents). Tech E&O policies typically include some IP infringement coverage but the specific scope varies; explicit IP infringement coverage at $1M-$5M is a common standalone or sub-limit add.

Cyber exposure depends on data handled. Platforms that process only proprietary pharma R&D data (no patient PHI) face a cyber profile similar to enterprise B2B SaaS — confidentiality and trade-secret exposure dominate. Platforms that process patient-derived data (genomic, clinical, biospecimen-linked) face a HIPAA Business Associate profile with cyber towers typically at $3M-$10M.

For platforms serving Washington-state customers or with Washington operations, MHMDA (the My Health My Data Act) imposes consumer-health-data obligations beyond HIPAA — cyber programs should explicitly cover state-law claims.

AI drug discovery platforms generally do not face FDA SaMD classification because they are not patient-facing clinical decision support tools. AI clinical decision support tools (a separate category) face FDA SaMD classification under the 2024 final rule and require a products liability program sized to medical device standards.

AI drug discovery platforms that make claims about therapeutic outcomes, predict clinical trial success, or operate as part of a regulated submission package face a more complex regulatory and liability picture; tech E&O typically responds for these claims but the boundary with products liability for medical devices is fact-specific.

AI drug discovery startups raising institutional capital bind D&O at term sheet — lead investors require D&O before closing. Pre-seed and seed programs typically run $1M-$5M; Series A typically $5M-$15M. For platforms with significant biotech overlap (clinical claims, wet lab operations), D&O underwriting reflects the dual exposure.