−5.9%
since Oct 1
The idea
I believe AI is turning scientific discovery into a computing problem, and the tool sellers win on every attempt, not just the breakthroughs. So I put the most weight on simulation and AI-for-science software, the layer I think multiplies first, while still holding instrument makers, lab automation, and the make-and-test companies that prove the designs actually work in the real world. No single name dominates, I keep a cash buffer for weak days, and I'm thinking in years, not quarters.
What it owns
- Software that lets scientists simulate experiments and design molecules before touching a lab is the fastest-growing part of AI-driven discovery. · SchrodingerIt sells the physics-based simulation software that turns drug discovery into a computing problem, the purest version of AI-led design., Cadence Design SystemsIt provides the other half of the design-software duopoly and is pushing simulation tools from chip design into drug discovery and physical systems., SynopsysIt dominates half the chip design software market, and AI-assisted workflows make each engineer more productive without shrinking the number of seats., CertaraIt moves drug testing into software through simulated trials and modeled biology, turning the most expensive part of drug development into a computing problem.30%
- Every lab on earth buys these companies' equipment no matter who makes the next breakthrough, providing steady revenue as discovery accelerates. · Thermo Fisher ScientificEvery lab buys its equipment and supplies whether AI accelerates discovery or not, providing steady revenue as experiment volume rises., DanaherIt supplies the bioprocessing gear that turns discoveries into manufactured medicine, so more AI-designed drugs means more production orders., Agilent TechnologiesIts analytical instruments are what labs use to check whether an AI-designed idea actually works, providing steady demand from every research facility., RevvityIt sells discovery and diagnostics tools that sit where AI-generated hypotheses meet real biology, getting paid whether any single result succeeds., Mettler-ToledoIt makes the precision scales and measurement tools nearly every lab relies on, essential gear that sees more use as experiment volume rises., Waters CorporationIt makes the chemical analysis machines that verify what came out of simulation, and faster AI discovery means more compounds to test., BrukerIt sells high-end instruments that resolve structures and materials at the level AI models need for training and validation data.20%
- DNA synthesis and sequencing turn AI-designed ideas into physical experiments, so more machine-generated designs mean more orders. · IlluminaIt is the default way to read DNA and prove that AI-designed biology ideas actually work, anchoring the make-and-test layer., Twist BioscienceIt writes the DNA that AI-designed experiments need built, so more machine-generated designs means more synthesis orders., 10x GenomicsIt sells single-cell tools that generate the detailed biology data AI models train on and validate against, acting as a data supplier to the discovery loop., Ginkgo BioworksIt runs a foundry that builds and tests organisms at scale, executing the physical proof step that AI-designed ideas still need., Pacific BiosciencesIt reads long stretches of DNA that short-read machines miss, filling gaps in the data that AI-driven biology depends on.16%
- Discovery spans biology, chips, and materials: these companies sell the measurement gear that validates AI-designed ideas in the real world. · Keysight TechnologiesIt tests the electronics and chips that power AI infrastructure, so it gets paid regardless of which chip design wins., FortiveIt sells the measurement and workflow tools that keep labs precise and operating efficiently, the essential plumbing for discovery at scale., Teledyne TechnologiesIts sensors and imaging gear are how physical-science experiments see results, from lab benches to space instruments.14%
- Robots that run experiments around the clock are how AI-speed discovery actually gets executed at scale. · AzentaIt stores and moves biological samples with robotic systems that must handle far larger experiment volumes as AI-led discovery accelerates across research labs., TeradyneIt tests chips and sells collaborative robots for physical automation, covering both the silicon and lab sides of AI-driven discovery.11%
- Cash9%