Platform vs Asset Archetype

The core business model question in AI-bio: should a techbio company be a platform (license access to technology, generate partnerships, de-risk through fee income) or an asset company (own drug candidates, go through trials, capture the full upside of approval)? Most early-stage companies tried to straddle both. The market in 2025–2026 is forcing a reckoning.

Platform archetype — Companies like Recursion, Schrodinger, AbCellera, and Ginkgo position their AI/computational platform as the product. Revenue comes from SaaS licensing (Schrodinger, Certara), research collaborations (AbCellera with Lilly, AbbVie, etc.), and milestone payments from pharma partners (Recursion's $20B+ in possible milestones). The risk: platform companies must prove their technology makes pharma more productive, a bar that requires clinical validation they don't fully control.

Asset archetype — Companies like Insilico Medicine, Iambic Therapeutics, and Genesis Therapeutics use AI discovery internally and own the resulting drug candidates. If the drug works, they capture far more value. The risk: they face the same Phase 2/3 attrition as any biotech; the AI discovery advantage only meaningfully compounds if Phase 2/3 rates improve (still unproven at scale).

Hybrid / spinout model (2024–2026) — Isomorphic Labs is the most interesting example: born from Google DeepMind's AlphaFold research, it operates both as a platform (pharma partnerships with Lilly and Novartis) and as an asset company (developing its own internal pipeline, targeting IND by end-2026). The $2.1B Series B implies investors believe Isomorphic can do both. Generate Biomedicines' February 2026 IPO ($400M, GENB) is another — Novartis partnership + own pipeline + public market.

The cautionary sub-ledger — Multiple AI-bio platform companies have underperformed. BenevolentAI's lead compound failed efficacy in 2023 (market cap collapsed from ~$1.5B to near-zero). Exscientia was acquired by Recursion in a distressed merger (Nov 2024, $688M — well below prior private valuation). The implicit promise that "AI will raise Phase 2 success rates" is untested at statistical scale.

Key Claims

• Pharma is hedging by partnering with multiple AI platforms simultaneously — Lilly has deals with Isomorphic, Iambic Therapeutics, Recursion; Novartis has deals with Isomorphic, Generate Biomedicines. Pharma's behavior signals that no single platform has proved sufficient. Evidence: strong (onhealthcare.tech)
• Platform companies have been de-rated relative to 2021 peaks — Recursion (RXRX) and Schrodinger (SDGR) are trading well below all-time highs despite continued clinical progress. The market is skeptical of platform multiples without clear clinical proof. Evidence: strong (public market data, 2024–2026)
• Asset companies with clinical results attract the largest deals — Insilico's Nature Medicine Phase IIa (Jun 2025) unlocked a $2.75B deal with Lilly; Genesis Therapeutics won Incyte + Lilly deals. Clinical proof drives deal value disproportionately. Evidence: strong (IntuitionLabs Lilly-Insilico analysis)
• IPO window for AI-native biotech opened in early 2026 — Generate ($400M, Feb 2026) and Eikon ($381M, Feb 2026) suggest public markets will accept the archetype. Priced as biotech hybrids, not pure-play software. Evidence: strong (BioPharma Dive)
• Isomorphic is the platform-thesis flagship — engine first (IsoDDE), assets second (17 programs) — IsoDDE is the productized engine; the company runs 17 active programs across oncology, immunology, and cardiovascular, targets first AI-designed drugs in Phase 1 by end-2026, and is funded by a $2.1B Thrive-led Series B. It is the clearest live test of the "do both" hybrid. Evidence: moderate (Isomorphic IsoDDE)
• The platform→asset bridge is still unproven: zero approved AI-designed drugs to date — Per the IsoDDE-context report, AI drug discovery drew more capital in 2024–2026 than any biotech subsector since immuno-oncology, yet has zero approved drugs. Recursion has spent >$1B over a decade without a commercial drug. The capital is betting on the bridge, not standing on it. Evidence: moderate (Isomorphic IsoDDE)

Open Questions

• Can a platform company like Recursion generate reliable milestone revenue from pharma partners while running its own clinical pipeline without running out of cash?
• Will Isomorphic prove the "both" strategy — platform + asset — or will investors eventually force a choice?
• At what Phase 2 success rate improvement does the AI-asset company model become clearly superior to traditional biotech?

Related Concepts

• The TechBio Thesis — the structural bet underlying both archetypes
• The Data Moat Question — platforms bet on data; asset companies don't need a moat to win
• Generative Biology — the technology both archetypes are built on

Changelog

• 2026-06-15 — Initial compilation; Isomorphic hybrid model, cautionary cases, IPO window covered
• 2026-06-24 — Compiled new sources (isomorphic-isodde-human-trials)