The recent discourse at the Mobile World Congress (MWC) in Barcelona suggests a fundamental pivot in the computing landscape. As Gustavo Fuentes noted for ABC, the event has long transcended its “mobile” nomenclature.
We are witnessing the birth of a new substrate for ubiquitous intelligence. For those in the engineering trenches, the signal is clear: we are moving from a “Mobile-First” era to an “Agentic-First” paradigm.
The Rise of the Agentic OS This shift is most visible in hardware announcements coming out of China. As reported by La Vanguardia, a new generation of smartphones is granting Large Language Models (LLMs) “deep access” to the device’s core functions.
Unlike sandboxed Western AI apps, these systems bypass traditional app-layer restrictions. This represents a move toward an “Agentic OS” where the model operates at the kernel or system-service level.
Hardware-Level “Thinking” The efficacy of this integration depends on the underlying “thinking” capabilities of the models. According to Xataka, the emergence of Qwen3-Max-Thinking has placed Alibaba’s models in direct competition with Google’s Gemini 3 Pro.
The “Thinking” suffix points to a significant architectural trend: – Implementation of advanced Chain-of-Thought (CoT) processing. – Increased test-time compute for proactive problem-solving. – Specialized NPU (Neural Processing Unit) acceleration to prevent thermal throttling.
From Digital Agents to Physical Embodiment The implications extend far beyond the screen. Omar Hatamleh, a veteran of NASA’s Goddard Space Flight Center, highlighted a future where this intelligence is embodied in humanoid forms.
Hatamleh posits an “Expertise-as-a-Service” model. Imagine domestic robots providing medical monitoring for a subscription fee of roughly 20 euros a month.
The Engineering Challenge The transition from a smartphone to a humanoid robot is a matter of scaling sensor fusion and real-time actuation. The technical bottleneck is no longer just the intelligence of the model, but the latency between perception and action.
As we move forward, we face a geopolitical split in privacy engineering. Can we build a competitive agentic experience that is “privacy-by-design,” or does truly helpful AI require the total surrender of the digital sandbox?
Source: https://www.abc.es/opinion/sevilla/gustavo-fuentes-despues-movil-20260312203834-nts.html
