Edge AI & IoT

Sector thesis

Edge AI & IoT is about running artificial intelligence directly on small devices—phones, sensors, cameras, factory equipment—rather than sending all data to a distant data center. Instead of your smart home camera uploading video to the cloud, it processes what it sees locally. This saves bandwidth, reduces latency (delay), and keeps sensitive data private. Why now? Three forces collide: AI models are getting smaller and faster (you don't need a supercomputer to recognize a face anymore), battery life and chip efficiency are improving, and companies want to avoid the cost and privacy risk of constantly uploading data. Factories want real-time defect detection. Hospitals want instant diagnostic alerts. Retailers want in-store analytics without cloud dependency. This is a structural shift, not a fad. The sector splits into three overlapping pieces: specialized chips (semiconductors designed for edge inference), software platforms (tools that compress and deploy AI models to devices), and connected devices themselves (industrial sensors, smart cameras, wearables). A chip maker might focus on ultra-low-power processors. A software company might sell tools to shrink a large AI model down to fit on a phone. A device maker might embed both into a new product. The biggest risk: this is still early. Many edge AI products are niche or experimental. Adoption timelines are unpredictable. Also, the AI compute sector (data centers, cloud inference) is already mature and well-funded—edge AI is smaller and more fragmented, so it's easier to pick the wrong company. For a retail portfolio: watch for companies shipping real products (not just prototypes), expanding into new industries, and growing revenue consistently. Look at gross margins (the percentage of each sale that's profit before overhead)—healthy edge AI businesses should show improving margins as they scale. This sector fits a growth portfolio, but size your position accordingly given the execution risk.