An AI Assistant for Everyone
Adl@adl05x
The idea
I believe AI assistants become a daily habit for almost everyone within a decade, across phones, laptops, glasses and work. Since the leading assistant labs are still private, I own the companies that win either way: the platforms distributing assistants to billions, the data-center chips and networking running every conversation, and the on-device chips bringing them to phones and wearables. The basket leans on large, profitable platforms for stability, with smaller bets on chips and infrastructure where growth could surprise, cash stays modest, and this is built to hold for years, not trade.
What it owns
- These companies own the phones, search, apps and clouds where people will actually meet their assistant, and several of them back the private AI labs. · Alphabet Inc.It can hand an AI assistant to billions of people through Android, Search and Chrome, so long as it stays a top-tier assistant built into the devices people already use daily., Microsoft CorporationIt puts an assistant inside Windows and Office where most office work happens, and its stake in OpenAI means it gets paid when ChatGPT grows, so long as Copilot stays in regular use at work., Apple Inc.The iPhone reaches over a billion people, so whatever assistant they talk to each day mostly arrives through Apple's hardware and software, as long as Apple controls which assistants sit on the iPhone., Amazon.com, Inc.It has Alexa in millions of homes, rents out the cloud computers that run assistants, and backs Anthropic, the maker of Claude, so long as its cloud keeps winning AI work., Meta Platforms, Inc.It can distribute an AI assistant across WhatsApp, Instagram and Facebook, and its glasses represent an early bet on wearable assistants, provided daily usage grows and AI spending eventually generates returns.40%
- Every question asked of an assistant runs on these chips, so more users means more chips sold. · NVIDIA CorporationIt makes most of the chips that run assistant conversations in data centers, so its position holds as long as it keeps the bulk of AI chip sales even as large customers design their own., Broadcom Inc.It builds custom AI chips for the biggest tech companies and the networking gear that links them, so it gets paid as those companies scale their own assistants, as long as they keep ordering from Broadcom., Taiwan Semiconductor ManufacturingNearly every advanced AI chip in this take gets made in its factories, so it holds as long as it stays the go-to manufacturer for leading-edge chips and its Taiwan operations stay intact., Advanced Micro Devices, Inc.It is the main alternative to Nvidia for data-center AI chips, and large buyers want a second supplier as assistant use grows, so long as it keeps landing major AI data-center deals., Micron Technology, Inc.AI chips need huge amounts of fast memory, and it is one of only a few companies that makes it, so assistant growth means more memory sold, as long as demand stays strong., Marvell Technology, Inc.It designs custom chips and the optical parts that move data between AI servers, which is plumbing every assistant needs, as long as cloud companies keep choosing it for custom chips.34%
- Assistants are moving onto phones, PCs and glasses, and these chip designers get paid for every device that runs AI locally. · Arm Holdings plcIt designs the chip blueprints inside almost every phone and many laptops, and collects a fee on each chip shipped, so long as on-device AI keeps building on its designs., Qualcomm IncorporatedFor assistants to run on phones, laptops and glasses without always calling the cloud, those devices need AI-capable processors, and it makes most of the ones outside Apple., Lattice Semiconductor CorporationIt makes small, low-power chips that let devices like glasses, cameras and sensors handle simple AI tasks without draining battery, betting assistants spread into wearables and edge devices.12%
- Billions of assistant conversations need fast networks and cooled, powered data centers, which these companies supply. · Arista Networks, Inc.It sells the high-speed switches that connect thousands of AI servers inside data centers, so more assistant traffic means more of its gear, as long as cloud companies keep buying for AI buildouts., Vertiv Holdings CoAI data centers need far more power and cooling than older ones, and it supplies that equipment, so this holds as long as data-center construction for AI keeps growing.7%
- Cash7%