Mobile World Congress (MWC) 2026 has opened in Barcelona with a more measured tone than previous years. If 2025 was about optimism and scaling GenAI pilots, day one of this year’s show suggests the industry is moving into a more complex phase: operationalising AI, defining ownership of intelligence, and defending its workforce impact.

AI is still the headline act. But the narrative has shifted from experimentation to control.

Agentic AI under scrutiny

Google used day one to directly address growing concerns that agentic AI could lead to large-scale job losses in the telecom sector.

Gemini Enterprise was positioned as a reasoning engine for higher levels of autonomous network operations. The focus, Google argued, is operational expenditure reduction and service quality improvement rather than headcount cuts.

Operators, including Deutsche Telekom and Vodafone, are already deploying these capabilities. Google cited work with Bell Canada that reduced wireless-related customer calls by 25%.

Technically, the stack is becoming more ambitious. BigQuery is automating telemetry data cleansing and tagging, while Spanner is supporting large-scale digital twins. The message is clear: networks are being re-architected to think, not just transport traffic.

Google also emphasised openness, noting that alternative LLMs such as Anthropic can be integrated. In practice, switching models in production environments remains complex, but the signal to operators was deliberate: flexibility matters.

Open Telco AI: Industry-specific intelligence

With conversations about AI spanning stages, booths, and advancements, it’s no surprise that it is once again a key pillar of the showfloor.

The GSMA also got in on the act by announcing the launch of Open Telco AI, a new dimension to the AI conversation.

The premise is that frontier models, whether from OpenAI, Google or Anthropic, are not optimised for telecom network data. Interpreting network telemetry, automating assurance, and orchestrating services require domain-specific training.

Open Telco AI will provide a shared portal for open telco-focused models, datasets, tools and compute. Founding supporters include AT&T and AMD, alongside operators such as Orange, SK Telecom, SoftBank and Vodafone.

The broader implication is strategic. Telecoms is seeking to shape its own AI layer rather than relying entirely on hyperscaler ecosystems.

Day one takeaways?

MWC 2026 has opened with a more mature AI debate. The focus is no longer on whether AI will transform telecoms. It is how, who controls it, and what the operational and workforce implications will be.

In 2026, AI must prove its economic value. Last year was about pilots and possibilities. This year, the question is simpler and more unforgiving. As TelcoDR / Totogi CEO Danielle Rios (DR) put it in her pre-MWC video, “Show me the money!”

The expectation is not theoretical efficiency gains, but measurable profit, revenue protection or margin improvement flowing to the bottom line. As operators move from experimentation to scaled deployment, AI is no longer judged on technical elegance alone. It is being assessed on whether it can navigate the complexity of real telco environments and deliver tangible financial outcomes. In the AI-native era, intelligence without economic impact will not be enough.

Alongside that, the infrastructure roadmap continues to evolve, increasingly framed through an AI-native lens.

Day two is likely to push that conversation further. Until then!