- Carriers building infrastructure to last decades should prioritize architectural trajectory over AI hype cycles
- Today’s hyperscaler AI is a powerful pattern prediction, but without grounding, causality or consequence, it isn’t true intelligence
- Artificial general intelligence is the north star for carriers – but it will arise not from larger LLMs, but from Industry 4.0 systems embedded in and interacting with the physical world
Today marks the opening of the GSMA’s megashow, Mobile World Congress 2026 in Barcelona. As you brace for the bombardment of AI-related announcements about to be fired at the telecom industry, here’s something to keep in mind:
Nothing good happens quickly in carrier networks. They’re built for resilience, determinism and five-nines reliability. Yet U.S. hyperscalers — including Google, AWS, and Microsoft — are working at hyperspeed to rewire the very DNA of global communications using probabilistic AI language models.
The truth is, hyperscaler AI isn’t what most people imagine AI to be. It’s a fusion of large-scale machine learning and probabilistic language prediction, wrapped in a chatbot interface cosplaying as sentience and intelligence.
But it isn’t connected to physical reality. It doesn’t perceive. It can’t test hypotheses against the world. It doesn’t operate on causal understanding. It has no intrinsic agency.
What it does do is generate coherent descriptions of reality — but without ever encountering it.
That’s important. Intelligence that never touches the world can simulate reasoning, but it cannot develop true situational awareness. It cannot experience the consequence of action or inaction. And without consequence, there is no genuine understanding — only mathy guesswork.
All those attributes should arrive with the next iteration of AI: artificial general intelligence (AGI), which doesn’t exist yet.
The simple fact is you can’t reach AGI simply by scaling up language models.
Scaling statistical pattern recognition is not the same as building general intelligence (because intelligence is not autocomplete at scale, and reality is not a word puzzle). And the fact that there is a perception in the industry that AGI could grow out of large language models says something about the moment we’re in — sitting alongside promises by Google DeepMind’s CEO that AI will cure all disease within a decade or by Nvidia that new compute architectures will magically solve their own energy demands.
What’s missing is that connection between statistical language fluency and real-world intelligence.
Real intelligence requires grounding. It requires the ability to see, hear, feel and move — to test theory against physical reality. It requires persistent memory of lived experience, not just tokens in a context window. It requires causal models built through interaction, not correlations extracted from text.
AGI will not emerge solely from larger language models. It will emerge from systems connected to the physical world — vast distributed networks of sensors, devices, and agents continuously interacting with their environments.
And guess what? This is where today’s AI intersects with Industry 4.0 – the global shift to industrial digitalization. It’s on that border — between cyber and kinetic — that the AGI breakthrough will occur. When AI algorithms don’t just describe reality but push and pull against it; when models don’t just predict outcomes but experience consequences; when AI at last closes the loop between sensing, reasoning and action.
Industry has already built the nervous system. AI will become the brain — not through hyperscaler chatbot intelligence, but through embodied, industrial-scale cognition.
The pieces are in place, but no one knows when they will converge into truly embodied, physically grounded AGI. But we do know this: within a few years, many of the AI implementations being hyped today by hyperscalers will look laughably primitive in hindsight.
And if you’re a carrier designing infrastructure meant to last decades, it may be worth stepping back. Think in longer timelines. Focus on trajectory, not hype cycles. Because the AI you’ll see this week in Barcelona is not the destination. It’s a prototype.
For more of this kind of thing, click here.
And check out all the FNTV coverage of Mobile World Congress 2026 at Fierce Network.
Stephen M. Saunders MBE is a communications analyst and USPTO-registered inventor examining how digital infrastructure — 5G, cloud, and AI — is reshaping industry, power and society, as well as underpinning the emerging, ubiquitous global digital economy. As anchor of FNTV and a longtime industry insider, he focuses less on growth narratives and more on execution, risk and how hyperscale technology is distorting markets, governance and society at scale.
Opinion pieces from industry experts, analysts or our editorial staff do not necessarily represent the opinions of Fierce Network.
Read all of our coverage from Mobile World Congress 2026 in Barcelona here.