"The artificial intelligence (AI) industry has often operated on a comfortable assumption: intelligence lives in the cloud. Progress, by that logic, flows from ever-larger data centers, ever-bigger models, and ever-more centralized compute controlled by hyperscalers. Training scale and inference scale have become shorthand for innovation itself. But at the CES 2026 conference in Las Vegas last week, Lenovo, the world's largest PC maker, challenged those assumptions directly."
"Rather than touting marginal gains in model size or benchmark scores, Lenovo used its Tech World keynote at Sphere to make a more structural claim about where AI is headed - and where it risks breaking down if the industry clings too tightly to the cloud. The real constraint in AI, executives argued, is no longer intelligence. It is deployment: where AI runs, who governs it, and how reliably it shows up in real-world settings."
""Our vision is to make AI user-centric, simple, and a single point of entry with a consistent experience," Yang Yuanqing, Lenovo's chairman and CEO, told me during an in-person interview at CES 26. "With hybrid AI, we give users access to general-purpose models in the cloud, highly specific models on devices, and models at the edge or on servers. It shouldn't matter where the models or agents sit.""
The cloud-centric assumption that intelligence resides only in large data centers is no longer sufficient. Deployment has emerged as the primary constraint: where AI runs, who governs it, and how reliably it functions in real settings. Hybrid AI is presented as necessary, agentic by design, and personal by default, operating across devices, private infrastructure, and public clouds. The cloud retains importance but no longer occupies the sole center of gravity. A strategy that spans AI-native PCs and smartphones, inference-optimized servers, agentic software platforms, and partnerships enables flexible model placement and a consistent, user-centric experience.
Read at Forbes
Unable to calculate read time
Collection
[
|
...
]