Everyone Gets an AI Assistant
Velislav Gerov@velislavgerov
The idea
I believe a personal AI assistant becomes as common as a smartphone, something everyone uses daily. To own that shift, I hold both sides: the big platforms whose assistants people actually talk to, and the chips that let those assistants run right on the device. The sharper bet is memory and processors, since on-device AI needs far more memory per phone and laptop. This plays out over years, so I keep the basket broad, lean on steady large companies, and hold a modest cash reserve.
What will be bought
- These companies operate the assistants people interact with daily, positioning them to benefit if assistants become essential on devices worldwide. · Alphabet Inc.Gemini and Android together reach most of the world's phone users, making Alphabet a default path for assistant adoption., Apple Inc.Apple places an assistant on every iPhone through Siri and builds AI to run directly on the device itself., Microsoft CorporationCopilot reaches office workers through Windows and workplace applications, making it the first assistant many encounter daily., Meta Platforms, Inc.Meta AI appears in its apps and glasses, offering a path for assistants to reach people through wearable devices., Amazon.com, Inc.Alexa already runs at scale in millions of homes, giving Amazon early experience operating a personal assistant in daily use.35%
- These chips power phones and laptops to run assistants locally, reducing dependence on cloud computing for everyday use. · Qualcomm IncorporatedQualcomm processors inside most Android phones enable assistants to run locally instead of in data centers., Arm Holdings plcArm collects a royalty on nearly every phone chip and an expanding share of laptop chips as AI adoption spreads., Advanced Micro Devices, Inc.AMD laptop processors include built-in AI engines that let assistants operate without connecting to the cloud.19%
- Chip designers and manufacturers are essential partners for every company building AI processors, regardless of which assistant wins. · Taiwan Semiconductor Manufacturing CompanyTSMC manufactures advanced chips for Apple, Qualcomm, AMD and most other companies in this take., ASML Holding N.V.ASML is the sole maker of machines required to produce the most advanced chips that power AI processors., Cadence Design Systems, Inc.Cadence design software is used to create every new AI chip for phones, laptops, earbuds and other devices.16%
- Running assistants on devices requires more memory per phone and laptop, driving demand for memory chips across the industry. · Micron Technology, Inc.Memory per device rises as assistants move onto phones and laptops, and Micron is one of three major suppliers., Sandisk CorporationFlash storage on devices grows as on-device assistants become larger and require more space to hold AI models locally.12%
- These specialized processors enable assistants to operate in earbuds, glasses, cameras and wearables beyond traditional phones and laptops. · Synaptics IncorporatedSynaptics makes low-power AI chips that enable assistants to function in smaller gadgets beyond phones., Lattice Semiconductor CorporationLattice produces always-on chips that listen and sense without draining batteries, essential for assistants in wearables., CEVA, Inc.CEVA licenses AI and wireless chip designs used in earbuds and wearables as assistants expand to smaller devices., Ambarella, Inc.Ambarella designs AI chips for cameras, which matters if assistants gain the ability to see through glasses and other devices.10%
- Cash8%