Biology Data Picks and Shovels
Yousif Al-Dujaili@yousifduj
· 1
−3.8%
since Oct 8
The idea
I believe AI models for biology are bottlenecked by experiment data, and that frontier labs and pharma will keep paying for it for years since no single company can run every experiment itself. This basket owns three layers: wet labs that generate and sell that data, the equipment and supply companies every lab depends on regardless of who owns it, and cell and tissue suppliers benefiting as testing shifts toward human tissue. I weighted it toward the steadier suppliers, kept smaller data-lab names sized down, held some cash, and built it to play out over five to ten years rather than two.
What it owns
- Every lab, whether built by frontier researchers or rented out, buys instruments and supplies from these companies. · Thermo Fisher ScientificIt sells instruments, reagents and plastics that nearly every lab buys, whether owned by frontier researchers, drugmakers or outside contractors., DanaherIt owns instrument and screening businesses that labs buy regardless of ownership, and has announced AI-powered autonomous labs for drug discovery., Agilent TechnologiesIt makes analytical and cell analysis instruments that labs use to measure what happened in an experiment., IlluminaIts sequencers produce much of the genetic data that AI models for biology train on, so more experiments means more sequencing demand., Waters CorporationIt makes instruments that measure molecules in drug testing, a step that nearly every lab repeats repeatedly., Twist BioscienceIt makes synthetic DNA that labs use to build and test thousands of designs, so it grows as experiments get larger in scale., RevvityIt sells high-volume screening and automation tools plus cell instruments needed to run thousands of experiments fast enough to feed AI models.46%
- These companies run experiments that AI labs and drug makers cannot run themselves, then get paid for the data. · Charles River LaboratoriesIt is the largest outsourced lab for preclinical drug testing when drugmakers and AI labs cannot run every experiment themselves., Tempus AIIt sells molecular and clinical data to drugmakers and has announced multi-year collaborations for data-driven drug development., Recursion PharmaceuticalsIt runs automated experiments at large scale and trains models on biology data, creating a data-for-AI loop that partners pay to access., Ginkgo BioworksIt rents out its automated wet lab to customers who pay to run experiments and generate data instead of building labs in-house., AzentaIt stores biological samples and runs sequencing as a service, getting paid when labs outsource work instead of doing everything themselves., AbsciIt pairs its own wet lab with AI to design antibodies, with partners paying for its lab work and design services.28%
- Testing drugs on lab-grown human tissue instead of animals requires stem cells, growth media and tools to measure what the cells are doing. · Bio-TechneIt sells stem cell media, growth factors and organoid tools that labs need to grow human tissue for drug testing instead of using animals., 10x GenomicsIt makes tools that read tissue cell by cell, generating rich data from human tissue models used in drug testing., BrukerIt makes imaging and measurement tools that examine tissue and molecules in detail, a key step when testing drugs on human tissue.17%
- Cash9%