1. What happened
Astex, Bristol Myers Squibb and Takeda joined the existing AbbVie and Johnson & Johnson initiative. Local computation permits model learning without pooling the underlying records.
2. Buyer / seller
AISB Federated OpenFold3 Initiative, AlQuraishi Lab, Apheris is the AI buyer or model-development side. AbbVie, Johnson & Johnson, Bristol Myers Squibb, Takeda, Astex Pharmaceuticals owns or contributes the proprietary corpus.
3. Economics
No data-specific consideration disclosed. Economics scope: not disclosed. Strategic Signal has not estimated undisclosed consideration.
4. Proprietary data
Proprietary experimentally determined protein–small molecule structures supplied for federated fine-tuning. This observation records the October 2025 consortium expansion. Historical depth: not_disclosed. Refreshability: continuing network; contribution schedule not disclosed.
5. Rights and restrictions
Local computation and contributor-defined access policies. Ownership: no disclosed transfer. Training: yes explicit. Post-training: yes explicit fine tuning.
6. Why the data matters for AI
Fine-tune OpenFold3 for protein–ligand prediction. Analysis: experimental structures directly constrain molecular predictions; they are not a complete history of clinical decisions or drug-development outcomes.
7. Asset Score — 87/100
Strong experimental grounding and learning fit. Structural results are not full decision histories. Unknown archive age and contribution cadence remain unscored. Score coverage: 84%.
| Factor | Weight | Rating / 5 | Points |
|---|---|---|---|
| Uniqueness / scarcity | 12 | 4.5 | 10.8 |
| Historical depth | 8 | Unknown | — |
| Decision → outcome richness | 13 | 3.5 | 9.1 |
| Decision Value Density | 13 | 4.5 | 11.7 |
| Real-world grounding | 10 | 5 | 10.0 |
| Domain economic value | 9 | 4.5 | 8.1 |
| Refreshability | 8 | Unknown | — |
| Proprietary advantage | 9 | 4.5 | 8.1 |
| Model-learning usefulness | 10 | 4.5 | 9.0 |
| Non-replicability | 4 | 4 | 3.2 |
| Rights usability | 4 | 4 | 3.2 |
8. Transaction Signal Score — 40/100
Clear learning arrangement, weak price discovery. Consortium-reported uplift is recorded separately and earns no independent-replication points. Score coverage: 85%.
| Factor | Weight | Rating / 5 | Points |
|---|---|---|---|
| Disclosed economics | 16 | 0 | 0.0 |
| Clean price discovery | 16 | 0 | 0.0 |
| Separability of data consideration | 15 | 0 | 0.0 |
| Explicit training / model-improvement use | 14 | 5 | 14.0 |
| Clarity of rights purchased | 12 | 4 | 9.6 |
| Competing bids | 10 | Unknown | — |
| Strategic buyer quality | 6 | 4 | 4.8 |
| Repeatability / precedent value | 6 | 4.5 | 5.4 |
| Independent model-uplift evidence | 5 | Unknown | — |
9. M&A / valuation implications
Analysis: rights to learn can be valuable without an asset sale. Reported predictive improvement is not a cash valuation or evidence of clinical benefit.
10. Evidence boundary
March 27, 2025 founding is context, not a second transaction here. September 14, 2026 performance evidence is a later update. Uplift is consortium-reported, not independently replicated; no price is disclosed.