−11.2%
since Oct 8
The idea
I believe AI is moving out of the cloud and onto the devices we already own, as open models shrink enough to run well on local chips. This basket spreads across the layers that benefit: device makers, edge silicon, memory and storage, and a small sleeve in the open-model ecosystem itself. I'm keeping it diversified rather than betting on one company, holding 10% cash to add on dips, and treating this as a two-year view rather than a permanent holding.
What it owns
- Companies selling devices with on-device AI capabilities, from phones to PCs to servers, where the take assumes inference stays local rather than moving to the cloud. · AppleIt designs and manufactures devices with dedicated neural silicon and has committed to running AI inference locally on iPhones and Macs rather than routing processing to the cloud., NvidiaIt produces the RTX consumer GPUs that have become the default hardware for people running open AI models on home computers and workstations., Dell TechnologiesIt sells corporate laptops and workstations that get replaced when businesses decide their fleets need AI processing capability, driving an upgrade cycle tied to local inference demand., HP Inc.It is one of the world's largest PC sellers, so an AI-driven device refresh cycle directly flows through its laptop volumes and pricing power., Super Micro ComputerIt builds compact servers that businesses deploy on-premises to run AI models locally, capturing demand for edge and on-site AI processing instead of cloud reliance., Corsair GamingIt sells memory, power supplies, and chassis components to people building personal computers for running AI models at home, representing the enthusiast end of local inference.33%
- Chipmakers designing processors, NPUs, and specialized silicon optimized for running AI inference on devices at the edge rather than in data centers. · QualcommIt manufactures Snapdragon chips with dedicated neural processors that handle on-device AI in phones and Windows-based AI laptops without relying on cloud processing., Advanced Micro DevicesIt ships Ryzen AI processors with built-in neural processing units into the AI laptop market, plus consumer graphics cards for local model inference., BroadcomIt supplies networking and connectivity chips that edge AI devices depend on, plus custom silicon design services for companies building proprietary AI processors., Arm HoldingsIt licenses chip designs that power most phones and a growing share of AI laptops, earning royalties as devices with its architecture increase in volume., Lattice SemiconductorIt makes low-power programmable chips for AI inference in cameras, laptops, and industrial equipment where edge processing is more efficient than larger processors., AmbarellaIt designs processors for computer vision AI inside cameras and vehicles, where inference runs directly on the device rather than in remote systems.27%
- Suppliers of DRAM, NAND flash, and storage drives that benefit from higher capacity and performance requirements as AI models run locally on more devices. · Micron TechnologyIt manufactures the DRAM that AI-capable devices require in materially higher quantities than previous generations as local model inference becomes standard., Western DigitalIt manufactures hard drives that store AI model files and the data they process on local machines and home setups as storage capacity per device grows., SandiskIt produces NAND flash memory for the SSDs that every AI-capable device ships with, and benefits as storage capacity requirements rise with local model adoption., Seagate TechnologyIt manufactures drives that store growing volumes of data created and consumed by local AI systems in homes and small edge servers near end users.19%
- Companies enabling the availability and manufacturing of open AI models and the hardware they run on, supporting the broader shift to local model deployment. · Meta PlatformsIt develops and releases Llama, an open-source AI model that enables capable inference to run on consumer hardware, making local AI practical and accessible., Taiwan SemiconductorIt manufactures nearly every AI-capable processor in this take, from Apple's neural chips to Qualcomm and AMD processors, and wins on volume regardless of which designer gains share.11%
- Cash10%