Climbing the AI Value Stack
Ethan@ethan
· 5
+18.7%
since Jul 25
The idea
I believe AI's value is climbing up the stack: the chips and cloud layers are built out and priced, so the next gains go to software infrastructure and then to AI-native apps, agents, and robotics where feedback loops compound fastest. The portfolio spans that whole ladder, from distribution-heavy incumbents to the data plumbing agents need to a smaller sleeve in robotics and consumer devices, with a cash buffer kept on the side. I'm weighting where value lands next rather than betting on one layer, and I'd rethink this if the migration upward stalls or reverses back toward the base layers.
What it owns
- Established software platforms with billions of users and deep enterprise data can deploy AI agents directly into existing products, turning their distribution advantage into immediate monetization. · Microsoft CorporationCopilot embedded across Office, GitHub, and Azure gives the company the widest enterprise funnel for deploying and monetizing AI agents as they mature., ServiceNowWorkflow automation is the natural environment for enterprise agents, and the platform sits atop the corporate processes that AI will increasingly execute end-to-end., SalesforceAgentforce converts proprietary customer data into deployed agents, positioning the company to shift pricing from per-seat to per-outcome as agents replace manual work., IntuitDecades of financial data in TurboTax and QuickBooks create prime real estate for autonomous assistants that can complete actual financial tasks for consumers and small businesses.26%
- Companies built around AI feedback loops from inception, where each user interaction improves the product and reinforces competitive moats. · Palantir TechnologiesIts AI Platform turns foundation models into deployed operational agents for enterprises and governments, demonstrating the clearest public example of application-layer monetization., AppLovinIts ad engine operates as a pure inference feedback loop where every transaction retrains the model, improving targeting and winning more spend, with margin concentration already visible., DuolingoAI directly lowers content costs and raises engagement, with the reinforcement loop between user behavior and model tuning embedded into the product itself., ShopifyAs shopping shifts toward AI agents transacting on behalf of consumers, the merchant platform becomes the commerce infrastructure backing the agent economy.22%
- AI agents require data pipelines, real-time monitoring, and edge infrastructure to operate at scale; these foundational layers benefit regardless of which applications succeed. · SnowflakeAgents depend on data access, and the company is where enterprise data lives, with AI layers that make it the query substrate for agent applications., DatadogEvery production AI application requires observability and monitoring, so the platform scales with application-layer growth regardless of which agents win in the market., CloudflareLow-latency agent inference requires edge infrastructure, and the global edge network positions it as the delivery layer for distributed AI applications.17%
- Autonomous machines and robots apply AI software to the physical world, where margins compound and data feedback loops create defensible advantages. · TeslaFull self-driving and Optimus robotics apply application-layer AI to the physical world, with real-time data feedback loops from millions of real-world interactions., Intuitive SurgicalThe established profitable robotics franchise benefits from decades of procedure data feeding increasingly autonomous surgical capability with entrenched market position., Global X Robotics & AI ETFBroad exposure to the robotics supply chain hedges single-name risk and captures upside across multiple physical AI players and subsystems.13%
- Devices in consumer hands, phones, glasses, and wearables, control the distribution channel through which AI assistants reach billions of people. · Meta PlatformsMeta AI reaches billions through existing apps while Ray-Ban glasses establish it as a leader in AI-native devices, creating a two-way bet on assistant adoption., AppleTwo billion devices under its control make it the default distribution channel for consumer AI assistants, establishing the endpoint where applications reach people globally.12%
- Cash10%