The 10-Billion-Device Thesis and the Post-Smartphone Landscape
Amon’s most provocative claim at Davos was a market sizing number: 10 billion devices. Not smartphones — AI devices. The argument runs through three steps.
First, the smartphone created the template: broadband cellular required a computer on the other end, so the phone became a pocket computer. 1.2 billion units are sold every year — the number one consumer electronics category.
Second, AI changes what counts as a device. If AI agents need to be with you all the time to be useful, then glasses, rings, bracelets, watches, earbuds, and ambient desk devices all become agent endpoints. Each is not extending a phone’s functionality — each is a standalone AI touchpoint.
Third, if everyone ends up with a watch, a ring, and glasses — each connected to an agent, not just a phone — you get order-of-magnitude device counts approaching or exceeding the smartphone installed base.
This reframes the entire addressable market for on-device AI silicon. The smartphone market is $500B+. A 10-billion-device personal AI market — even at lower ASPs per device — represents a TAM expansion that dwarfs the current handset opportunity. And Qualcomm’s Hexagon NPU is the only silicon architecture deployed at 2B+ devices today.
Fashion Constrains Form Factor, Not Compute
The Davos conversation surfaced a critical insight: “Humans already decided what they’re going to wear a long time ago.” This means the personal AI device revolution will be constrained by fashion and form factor preferences, not just compute capability. Glasses, watches, rings, earbuds — these form factors already exist. The question is which company provides the AI silicon that makes them intelligent.
Horizontal Platform vs. Vertical Integration
Amon made a prediction with significant implications: “I think the horizontal model is going to win versus the vertical model” in personal AI devices.
His reasoning: unlike phones, which most people carry, wearables intersect with fashion. People want different form factors, colors, brands, and styles. Different brands for different people, age groups, and preferences. “I think we’re going to see a lot of brands in wearables.”
Brand override (vertical):Apple’s approach — control the hardware, the software, the model, and the ecosystem. Premium pricing, closed ecosystem. This works for a segment but limits total addressable volume.
Technical excellence (horizontal): Qualcomm’s approach — provide the best silicon platform and let hundreds of device makers build diverse form factors. Meta for glasses, Samsung for XR, ECARX for automotive, hundreds of IoT OEMs for industrial.
The smartphone era was won by the vertical model (Apple captured most profits) and the horizontal model (Android/Qualcomm captured most volume). The personal AI device era may invert this: if diversity of form factors matters more than ecosystem lock-in, the horizontal platform captures both volume and value.
This is the same dynamic that plays out in the data center. NVIDIA’s vertical stack works brilliantly for training. But inference — which increasingly requires heterogeneous architectures and diverse deployment environments — may favor a more horizontal approach.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
