The Last EASY Trade Before It Explodes

BWB - Business With Brian•Jun 30, 2026

Summary

Brian discusses the emergence of Edge AI as the next critical phase in artificial intelligence, following the initial build-out of large data centers. He emphasizes that while data centers are essential for training AI, the real-world application will increasingly occur on devices at the "edge," such as phones, cars, cameras, and robots. This shift promises faster, more private, and cloud-independent intelligence, with the edge AI chip market projected to grow over 30% annually, surpassing data center growth. Brian notes that recent market pullbacks have created attractive buying opportunities for investors. He then details several companies positioned to lead this transition:

**Calix:** Brian explains that Calix is a key player at the internet's edge, building software platforms for home Wi-Fi gateways used by internet providers. The company successfully transformed its business model from a box seller to a recurring revenue software platform, leading to a doubling of revenue and a significant increase in gross margin to near 57%. Calix, which once lost money for a decade, is now quietly compounding its growth on the network edge by applying an AI layer to its platforms.

**Nvidia:** While widely known for powering AI data centers, Brian highlights Nvidia's relevance in Edge AI through its Jetson Thor, an "on-robot brain" designed for autonomous machines and robotics. Coupled with its Isaac software and Groot foundation models, Nvidia is enabling machines to see and move intelligently. He points to Nvidia's remarkable financial performance, with annual profit multiplying over 40 times in six years and gross margins exceeding 70%, underscoring its continued leadership in the broader AI space with robotics as a next growth engine.

**Ambarella:** Brian identifies Ambarella as a pure-play investment in edge AI vision, with approximately 80% of its revenue genuinely coming from this sector. Their chips enable cameras, drones, and robots to process visual information and make decisions directly on the device without cloud intervention. Ambarella has shipped over 46 million vision chips and recently secured a decade-long agreement with Hanwha. Although currently investing heavily for growth and not yet generating large profits, its revenue re-accelerated by about 37% in the past year, with operating losses quickly narrowing, making it a forward-looking investment for edge vision.

**Renesas:** This company, the world's largest maker of microcontrollers for automobiles, is now integrating real on-device AI into its silicon. Brian mentions their R-Car X5H, the first automotive chip on a 3-nanometer process, which delivers up to 400 trillion AI operations per second with significant power reduction. Major auto parts players like Bosch and Zephyr are already designing with it. Renesas maintains high-teens operating margins and has strategically acquired Altium and Pictorus to own the entire design workflow. Despite a recent one-time write-down, its underlying business remains profitable, though Brian notes it trades over-the-counter in the U.S., making it harder to buy.

**Broadcom:** Known for its custom AI chips (ASICs) in data centers, Brian points out Broadcom's leadership in this field, designing around 60% of the world's custom AI chips for major cloud providers. The company has seen its profit increase eightfold and operating margin climb from 17% to 40% over the last five years. For edge AI, Broadcom is embedding dedicated AI engines into Wi-Fi routers and home gateways, exemplified by its first Wi-Fi 8 chip with a neural engine, and a 5G platform with Samsung, positioning it at the forefront of the network edge.

**Mobileye:** Brian describes Mobileye as the purest way to invest in automotive edge vision, as its IQ chip acts as the "eyes and reflexes" for cars, processing camera frames directly within the vehicle. With IQ chips in over 230 million vehicles, Mobileye boasts one of the largest deployments of AI hardware globally. After overcoming a significant one-time write-down, revenue is climbing again, up 27% recently. Mobileye is also expanding into operating its own robotaxi fleet. Brian acknowledges the structural risk of Intel's large share block but emphasizes Mobileye's massive installed base as a strong competitive moat.

**Cohu:** This company, often overlooked, is crucial for ensuring the reliability of edge chips. Brian explains that Cohu provides testing equipment for every processor, sensor, and automotive silicon before it ships. He notes that Cohu is a cyclical business currently catching an upturn, with orders jumping 57% year-over-year, driven by the surge in edge and automotive silicon. Its recurring revenue from spares and services now constitutes nearly 60% of its business, and its Eclipse platform is gaining traction for testing high-performance AI processors, extending its reach beyond automotive into general compute.