Century Report arc
Pre-Competitive Data Cooperation in Science & Industry
When the cost of individual failure outpaces the value of competitive secrecy, rivals are pooling the foundational substrate of their industries (data, methods, baseline knowledge), treating it as shared infrastructure rather than proprietary moat.
- Status
- Active
- Developments
- 6
- Last 30 days
- 1
- Last covered
Current synthesis
Where it stands
Biohub's virtual-cell coalition establishes scheduled public release rather than permanent consortium exclusivity: government-funded work is unrestricted, while commercial data opens after one year (10-08). The five-firm structural-biology model remains inside its consortium (09-16), alongside HCMI's public cancer-model archive and shared cyber-incident efforts (08-06). The contest now includes an explicit alternative to permanent secrecy, though corporate funders still receive an exclusive head start.
Complete record
Development timeline
6 qualifying developments, newest first
October 2026
-
[DD+POD] Biohub's $1.8 billion coalition plans unrestricted government-funded work and public release of commercially funded cell data after one-year embargoes; DeepMind, Meta and Isomorphic Labs contribute $300 million.
September 2026
-
[DD+POD] Five drug companies' AI Structural Biology Network fine-tuned an open protein-folding model on 20,167 pooled proprietary protein-ligand structures via federated learning, outperforming public and single-firm versions, though the resulting model stays inside the consortium.
August 2026
-
[DD+POD] The publicly funded Human Cancer Models Initiative deposited 665 patient-derived models spanning 25 cancers into an ATCC archive available worldwide, including 153 rare-cancer models and 71 from non-European donors.
-
[DD+POD] Nvidia’s 120-company Open Secure AI Alliance proposed blame-free shared incident reporting through SAFE, adapting aviation’s near-miss model to agentic-security failures.
May 2026
-
Three independent labs (Wake Forest axolotl, Duke mouse, Wisconsin zebrafish) ran coordinated cross-species CRISPR/gene-therapy experiment and co-published in PNAS; lab-by-organism specialization had made conserved SP6/SP8 regeneration program invisible inside any single lab.
February 2026
-
Apheris ADMET Network: five pharma companies (Lundbeck, Orion, Recursion, Servier + one undisclosed) pool 80% of proprietary ADMET data via federated learning; targets 40-45% clinical trial failure rate from poor absorption/toxicity prediction; no proprietary data leaves secure environments: competitive dynamics yielding to shared-infrastructure logic when failure cost is high enough.