Own Your Context
Joshua Stylman@jstylman
−10.3%
since Oct 2
The idea
I believe AI models themselves are becoming a commodity, easy to rent, while the real lasting value is in owning your own history, notes and context. So instead of betting on the labs, this basket owns the picks and shovels: memory and storage makers, neutral software that holds data and files without tying it to one model, and a lighter slice of the chips running AI on phones and laptops. It leans hardest into storage and that neutral data layer, keeps steadier large companies as ballast, and holds 10% cash to add on weakness.
What it owns
- The take requires software platforms that store data and context independently of any single AI model, keeping information portable across different AI providers. · DropboxDropbox stores files and context without owning an AI model, functioning as a neutral memory layer that works across different AI providers rather than locking users to one lab., BoxBox holds enterprise content and integrates with any AI model without locking companies into a single vendor, keeping context portable instead of captive to one provider., MongoDBIt is the database businesses run their own applications and context on with no AI lab in the middle, so it gains importance if owned data becomes more valuable as models become interchangeable., CloudflareIt runs the infrastructure that lets companies serve data and AI without being trapped in one provider's world, the plumbing of portability this take depends on., ElasticElastic lets companies search and feed their own data into whatever AI model they choose, keeping context independent and portable rather than locked within one vendor's system., SnowflakeIt is where big companies keep their data and now run AI against it without giving that context to a model maker, working as long as enterprises treat their data as the asset and models as rented.30%
- Every device running AI locally needs far more memory and flash storage than models in remote datacenters, so the companies supplying it benefit as AI moves onto devices people and businesses own. · Micron TechnologyMicron makes the DRAM that local AI models load into, so demand for its memory grows if computing shifts toward on-device and local AI instead of pure remote data center models., SandiskSanDisk makes the flash storage inside phones and laptops where personal data and context physically live if kept on personal devices rather than in remote servers., NetAppIt sells storage to companies that want their data on their own terms wherever it lives, so it benefits if businesses keep their context in-house rather than hand it to a model provider., Seagate TechnologySeagate makes large, affordable drives that make archiving years of personal and business data economical, supporting the premise that ownership of one's own context matters., Western DigitalWestern Digital manufactures drives people buy for personal archives and small business storage, fitting the take if demand for owning and controlling one's own history grows.26%
- These chips run AI models directly on phones and laptops instead of sending data to distant servers. · QualcommQualcomm's chips run AI directly on phones, keeping data on devices and processing local instead of sending information to remote servers., AppleApple has built AI processing directly into a billion devices, betting on private on-device computation as the core experience rather than renting intelligence from remote labs., Advanced Micro DevicesIts processors power the AI PCs that can run models locally on hardware people own, so it benefits if the AI PC upgrade cycle happens and local models stay worth running., Arm HoldingsNearly every chip running AI on phones and small devices is built on its designs, so it collects a toll on the whole on-device shift without picking a winner.19%
- These larger, steadier companies build the machines and chips that let businesses run AI on their own premises with their own data, without renting from a lab. · Taiwan SemiconductorIt manufactures the chips for nearly every company in this take, so it wins whichever silicon maker does and provides steadier ballast for the multi-year shift of AI onto devices., Dell TechnologiesDell sells the servers companies deploy to run AI on premises with their own data, enabling in-house computing rather than renting models and memory from external labs., Hewlett Packard EnterpriseIts private cloud lets companies run AI where their context lives instead of shipping it to a lab, steadier ballast for the keep-your-own-data trend if businesses keep choosing private setups.15%
- Cash10%