The Tectonic Shift
alejo miragaya@miralejo
The idea
I believe AI is less a bubble than a tectonic shift, the global economy reorganizing around a new kind of work over many years. So this basket spreads across every layer of that shift: the chips and networking doing the computing, the power, grid and copper the buildout physically needs, the platforms renting AI out, and established companies turning more productive by using it. It leans on large, profitable names as ballast, keeps a modest cash reserve to add on weakness, and does not depend on any single layer being right, only on the economy continuing to rebuild itself around AI.
What it owns
- Every AI system runs on chips and networks, so the companies making them get paid regardless of which AI application succeeds. · NVIDIA CorporationIt sells most of the chips doing AI computing, so it gets paid as long as big buyers keep raising AI chip spending and it keeps most of that market., Taiwan Semiconductor ManufacturingAlmost every advanced AI chip gets made in its factories regardless of the chip's designer, so it gets paid whichever design wins., Broadcom Inc.It builds custom AI chips for big platforms and the networking gear linking thousands of chips together, a second way to own the computing layer., ASML HoldingIt is the only company making the machines needed to print the most advanced chips, so more AI chips mean more orders for its equipment., Micron TechnologyAI chips need huge amounts of fast memory, and it is one of only a few companies making it at scale., Arista NetworksIt sells the switches moving data between computers inside AI data centers, growing as those facilities get larger.31%
- Data centers need enormous amounts of electricity and metal, and those physical constraints are where shortages and pricing power emerge. · GE VernovaIt makes the gas turbines and grid equipment that produce and move the electricity data centers need., Vertiv HoldingsIt sells the cooling and power systems that keep AI data centers running as they generate intense heat., Freeport-McMoRanPower lines, data centers and grid upgrades all need copper, and it is one of the world's largest copper miners., Eaton CorporationIt makes the electrical equipment like switchgear and transformers that every new data center and grid upgrade needs., Constellation EnergyIt runs the largest fleet of US nuclear plants providing steady carbon-free power that data center builders want to buy for decades., Cameco CorporationIf nuclear plants become a major power source for AI, they need uranium fuel and it is one of the largest suppliers.25%
- The largest cloud and software companies rent AI to the rest of the economy and have the cash reserves to keep building. · Microsoft CorporationIt rents AI to businesses through its cloud and builds it into office software most companies already pay for., Alphabet Inc.It builds its own AI models and chips and rents both through its cloud, while search and video advertising fund the investment., Amazon.com Inc.It runs the largest cloud business, so much of the world's AI computing gets rented from its data centers., Meta PlatformsIt uses AI to pick which ads and posts each person sees, making ads work better and sell for more money.24%
- Established companies outside tech use AI to cut costs, showing the shift spreading through the wider economy. · JPMorgan ChaseIt can use AI to cut paperwork and costs across banking operations, letting it maintain strong profits while deploying the technology., Caterpillar Inc.It sells backup generators and engines data centers need, plus the heavy machines that build them and mine copper., Walmart Inc.It uses AI to run stores, inventory and delivery more cheaply, showing how the productivity gains spread beyond tech.12%
- Cash8%