−6.0%
since Oct 8
The idea
I believe we're shifting from paying for what you already know to paying for finding what nobody knows yet, and AI is accelerating that shift. So I own the toolmakers who profit from every attempt at discovery: instrument giants, simulation and AI-for-science software, DNA sequencing and synthesis, physical-science instruments, and lab automation. I've leaned harder into AI-for-science by adding a small basket of AI-native drug hunters running their own pipelines, funded by trimming everything else a touch and capped around 11% split across names since any one can swing hard on a single trial.
What it owns
- These companies sell the equipment and supplies that laboratories and drugmakers purchase regardless of whether each experiment succeeds, so they earn money on every attempt at discovery. · Thermo Fisher ScientificSells the equipment, consumables and outsourced trial work that nearly every laboratory and drugmaker purchases, so it earns money on each attempt at discovery whether or not that attempt finds anything., DanaherMakes the tools and bioprocessing gear that laboratories and drugmakers use to study biology and manufacture new medicines, so more discovery attempts mean more orders., Agilent TechnologiesInstruments measure what is in a sample, a step almost every chemistry and biology experiment needs, with consumables purchased on each run., Waters CorporationMakes machines that separate and identify molecules, used across drug research and testing, and earns repeat revenue from supplies and service on each run., BrukerBuilds high-end instruments that let researchers see molecules and proteins in new detail, the kind of gear used at the frontier of discovery., RevvitySells reagents, instruments and software that laboratories use for life-science research and diagnostics, much of it purchased again with every experiment., Mettler-ToledoPrecision balances and laboratory instruments sit on benches in research and manufacturing labs everywhere, capturing a small piece of a vast number of experiments.30%
- Software lets scientists and engineers test ideas on a computer before running real experiments, compressing the cycle between concept and physical testing. · SchrodingerSells physics-based molecular simulation software to most of the largest drugmakers, letting them test drug ideas on a computer before making them in the laboratory., SynopsysMakes the software used to design and simulate chips, including the artificial intelligence chips that power faster discovery, and is pushing that simulation know-how into physical engineering., Cadence Design SystemsSells chip design and simulation software and has been moving into molecular simulation for drug research, so it gets paid as both chips and molecules are designed on computers first., CertaraBuilds software that models how a drug will behave in the body, helping drugmakers and regulators cut down on trial and error during development.17%
- Reading and writing DNA is the foundational work of modern biology, and these companies sell the tools and services that let researchers and AI systems turn designs into physical molecules. · IlluminaMachines read DNA and earn money on the supplies used in every sequencing run, so more biology experiments mean more revenue from the leading platform., Twist BioscienceWrites custom DNA that researchers and artificial intelligence drug designers order to test their ideas in the laboratory, earning money on each design that leaves the computer., 10x GenomicsSells tools that look at genes one cell at a time, producing the detailed data that artificial intelligence biology models learn from and depend on., Ginkgo BioworksRuns automated laboratories and sells cell engineering and laboratory automation to others, providing make-and-test services for biology customers., Pacific BiosciencesMakes long-read sequencers that catch parts of the genome that shorter reads miss, a complementary technology to established sequencing platforms.14%
- Measurement and testing gear used in physics, electronics and materials research gets deployed on each round of experiments as engineers and researchers validate new designs. · Keysight TechnologiesTest and measurement tools are how engineers check new electronics, wireless and chip designs actually work, so it gets paid on each round of physical-science experiments., Teledyne TechnologiesMakes cameras, sensors and scientific instruments used to see and measure things in research, space and industry, serving a broad base of experimental work., FortiveSells measurement and testing tools that engineers and technicians use to check physical systems, a broad piece of the physical-science toolkit.11%
- These companies use AI to discover their own drugs rather than selling tools to others, placing a direct bet on which team actually makes the next breakthrough. · Recursion PharmaceuticalsRuns its own drug pipeline using automated labs and artificial intelligence to discover new medicines, the most direct bet on artificial intelligence hunting for drugs itself., Relay TherapeuticsUses computer models of how proteins move to design drugs, then takes them into its own clinical trials, betting that the design method produces better results than traditional approaches., AbsciUses generative artificial intelligence to design antibodies and tests them in its own laboratories, both for partners and its own pipeline, betting on a new method for drug discovery.11%
- Robots and automated storage systems let laboratories run far more experiments with the same staff, multiplying throughput and the attached revenue streams. · AzentaStores and manages biological samples with automated systems, providing the back-end infrastructure that more experiments depend on as labs automate sample handling., TeradyneMakes chip test systems and collaborative robots that automate laboratory and factory work, so it is paid as both chip output and automation grow.8%
- Cash9%