Commercialization
What Scientific AI Companies Sell Next: Experimental Data, Labs, and Drug Discovery Assets
Experimental services, laboratory infrastructure and drug assets: three ways scientific AI businesses earn returns.
WEEKLY 003 · EDITION RECORD
Testing predictions in experiments, clinical care and business.
Readings on scientific AI businesses, clinical AI oversight, neural grafts and antidepressant selection.
Contents
4 readingsR&D Allocation
Who Will Build the Systems to Oversee Clinical AI?
Patient support, supervision, implementation and independent evaluation: the research roles funded by ADVOCATE.
Frontier Science
How Far Can Human Neural Tissue Compensate for Lost Function?
Beyond graft survival: what could the animals do, and what remained impaired? Comparisons with dopamine and inhibitory-cell graft studies.
Frontier Science
Can measuring brain function help us choose an antidepressant?
How far did brain imaging, behavior and clinical information take medication selection? What SMART-D leaves to test next.
This issue’s insight: If AI Can Solve It, What Do You Sell? →
Topic states at issuance
Memory resilience
Compare sleep, representational drift and selective synaptic perturbation to investigate changes compatible with memory.
- Next evidence
- Matched perturbations with different targets, fixed versus adaptive readout, and retention with or without relearning.
Shield: screening adoption
Testing and revenue have grown. Track previously unscreened participants, diagnostic follow-up and assay improvements separately.
- Next evidence
- Who is newly screened, colonoscopy completion, testing costs and lesion-specific sensitivity.
Therapy & companion diagnostics
Linked approval; uptake and payment open
- Next evidence
- Confirmatory evidence, payer policies covering the new use, and clinical adoption.
Compute for science
A collaboration route exists from Japan. Check funding, compute allocation, disclosure and IP conditions.
- Next evidence
- Check eligibility, collaborators, compute allocations, publication and IP terms. Deadline: December 1 at noon ET; decisions scheduled for March 1, 2027.
AI–scientist collaboration
Institutional report; independent verification open
- Next evidence
- Independent replication, public evaluation methods, and records of successes and failures from hypothesis to experiment.
What Scientific AI Companies Sell Next: Experimental Data, Labs, and Drug Discovery Assets
Ginkgo announced experimental-services and in-house laboratory partnerships. Schrödinger contributed two drug programs to Tectora.
- Next evidence
- Repeat orders, laboratory operation and recurring revenue, asset rights and development costs.
Who Will Build the Systems to Oversee Clinical AI?
ARPA-H announced ADVOCATE development plans covering patient-facing AI, supervision, implementation and independent evaluation.
- Next evidence
- Missed errors, false alarms, appropriate clinical handoffs and evaluation in routine care.
How Far Can Human Neural Tissue Compensate for Lost Function?
A study examined human cortical organoid growth, connections, activity and behavior in mice with major developmental depletion of cortical and hippocampal excitatory neurons.
- Next evidence
- Manipulate graft activity during specific tasks to distinguish modulation of host circuits from information processing.
Can measuring brain function help us choose an antidepressant?
In SMART-D’s main analysis of 47 participants over eight weeks, selecting a marker-consistent drug did not demonstrate better outcomes.
- Next evidence
- Drug comparisons at the same treatment stage, and separate validation of early changes to inform subsequent decisions.
Record format & provenance
The archive stores complete article text, public evidence references, event dates, scheduled/observed distinctions, topic states and disclosed series. It contains no mailbox identifiers or private email bodies.
Editorial record window: 2026-09-23 — 2026-09-23 / Method: editorial-events-v1