Private AI, Paid by Agents
dhtal@dhtal
−7.4%
since Oct 14
The idea
I believe AI is going to push up against real-world limits over the next decade: power constraints, demand for privacy, and a shift toward agents as the way people buy things online. So this basket spreads across four layers: companies building out power generation and grid capacity, the servers businesses run on their own premises, products that sell privacy to everyday consumers, and the crypto rails that route agent payments in stablecoins, favoring the companies behind that infrastructure over the coins themselves.
What it owns
- Electricity and the infrastructure to deliver and manage it will become the primary constraint on AI expansion, benefiting both power generators and the companies that supply distribution and cooling equipment. · Constellation EnergyIt operates the largest nuclear fleet in the United States, providing the reliable baseload power that data centers require for continuous AI operations., GE VernovaIt manufactures gas turbines and electrical equipment that utilities and data center operators need to expand power capacity, with a large order backlog., EatonIt supplies the electrical distribution equipment connecting power grids to server racks in both large data centers and private corporate installations., VistraIt owns power generation assets in major US markets and benefits directly from increased electricity demand as AI workloads grow., VertivIt manufactures the power delivery and cooling systems that data centers and private installations require to operate AI hardware., NextEra EnergyIt operates regulated utilities and builds new power capacity, providing steady revenue growth as electricity demand increases over the long term.33%
- Organizations handling sensitive data and AI workloads prefer to keep those systems on hardware they control in their own facilities rather than relying entirely on cloud providers. · Dell TechnologiesIt manufactures AI servers for companies building private data center infrastructure, benefiting from both rising demand for on-premises systems and higher component costs., NutanixIt provides software that enables companies to run their own data centers as private clouds, capturing revenue when businesses move AI workloads in-house., Hewlett Packard EnterpriseIt sells servers, storage and private cloud infrastructure to organizations keeping sensitive AI systems within their own facilities., Cisco SystemsIt provides the networking and security equipment that companies need to connect and protect their own AI infrastructure inside corporate facilities.23%
- Consumers increasingly prefer products with privacy built in rather than adding it themselves, creating revenue opportunities for companies offering integrated privacy features and subscriptions. · AppleIt sells privacy as a core product feature, running AI models on devices themselves, and is expanding this approach into new consumer categories., CloudflareIt sits between websites and users, allowing it to manage how AI agents access content and can charge for that access while protecting user privacy., Gen DigitalIt sells privacy and security subscriptions to consumers through well-known brands, offering built-in protection without requiring technical setup., QualcommIt designs chips that enable phones and laptops to run AI models locally on the device, eliminating the need to send user data to external servers.22%
- As AI agents conduct transactions in stablecoins over blockchain networks, the platforms and currency issuers that facilitate those payments capture value from both transaction routing and from the stablecoins themselves. · Coinbase GlobalIt routes payments for AI agents through blockchain networks and earns a share of stablecoin transaction activity as agent payments grow., Circle Internet GroupIt issues the stablecoin used in most agent payments and earns interest on the capital backing every unit in circulation as adoption grows.15%
- Cash7%