The next phase in artificial intelligence could be “edge AI,” which lowers costs and improves security by embedding AI capabilities directly in smartphones and other devices.----- Transcript -----Welcome to Thoughts on the Market. I'm Ed Stanley, Morgan Stanley's Head of Thematic Research in Europe. And along with my colleagues, bringing you a variety of perspectives, today I'll discuss Edge A.I. It's Monday, the 20th of November at 2 p.m. in London. The last year has seen a surge in adoption of artificial intelligence, particularly for foundational model builders and consumer-facing chatbots. But we think the next big wave of A.I will be embedded in consumer devices, this is smartphones, notebooks, wearables, drones and autos, amongst others. Enter Edge A.I. This means running A.I algorithms locally rather than in centralized cloud computing facilities in order to power the killer apps of the A.I age. Generative A.I., cloud computing, GPUs and hyperscalers, that is, the large cloud service providers that run computing and storage for enterprises. They all remain central to the secular machine learning trend. However, as A.I continues to permeate through all aspects of consumer life and enterprise productivity, it will push workloads to hardware devices at the edge of networks. The US data firm Gartner estimates that by 2025, half of enterprise data will be created at the Edge, across billions of battery powered devices. The key benefits of A.I computation performed at the Edge are lower cost, lower latency personalization and importantly, higher security or privacy relative to centralized cloud computing. And the prize in moving these workloads to the Edge is large, we're talking some 30 billion devices by the end of the decade, but the hurdles are also significant. We think 2024 will be a catalyst year for this theme. And the companies that could benefit range from household name hardware vendors to key components suppliers around the world. But just as there are benefits to Edge A.I, there are constraints as well. Not all Edge devices are created equal, for example. The clearest limitations across hardware media are battery life and power consumption, processing capabilities and memory, as well as form factor, i.e. how they look. For example, mass market smartphones and notebooks today don't have the battery life or processing capability to run inferencing of the largest large language models. This will have to change over time, which will require investment predominantly in advanced proprietary silicon or custom ASICs as they're known, of which we've seen a number of announcements from big tech companies in recent weeks. The hardware arms race is really heating up in our view. It's important to note, though, that generative A.I. and Edge A.I are not mutually exclusive. In fact, Generative A.I. has reinforced the already growing need for edge A.I. Our consumer and investor trend analysis suggests that the theme is already moving into its upswing phase. Moreover, a slate of new product releases as soon as Q1 2024, such as Edge A.I enabled smartphones with embedded custom silicon, should drive further investor interest in this theme over the coming 12 months. And we think smartphones stand the best chance of breaking the bottleneck soonest and they also have the largest total addressable market potential in the short and medium term. This is an uncrowded theme which we think is in pole position for 2024. Thanks for listening. If you enjoy the show, please leave a review on Apple Podcasts and shared Thoughts on the Market with a friend or a colleague today.