This is The Road to Autonomous Networks.
Every month, we share the signals that tell us where autonomous networks are actually going.
You can subscribe on Linkedin here.
The signals from this month, below, point to the same place:
- Agents are ready for telecom, but most telco data layers are not ready for them.
- Connecting agents at a stack designed for dashboards just gives you... dashboards with a chat interface.
We are showing what an agentic data layer looks like at DTW Ignite next month.
If you are going, come and see it live.
Cheers.
— Vicent R. Soler Ruiz Co-founder & CEO, Kenmei
McKinsey: AI in network operations cuts total OPEX by 15–30%
Operators that have already scaled AI in ops report troubleshooting tickets down 30–70%, NOC costs down 55–80%, total OPEX down 15–30%.
Separately, McKinsey quantifies that a single network incident raises churn probability up to fivefold per customer.
The "AI in ops" conversation has moved from pilots to capex reviews.
Accenture: 79% of operators are still at autonomy Level 0 or 1
Only 22% target Level 4 by 2030. Just 10% expect to reach Level 5.
CTOs name legacy platforms (47%) and the AI skills gap (49%) as the main blockers.
The deeper issue isn't AI itself. Networks generate more data than ever, but decision speed hasn't moved.
That gap is where autonomy stalls.
Telefónica, T-Mobile, Deutsche Telekom: agents are now in production
- Telefónica: 12 use cases certified at Level 4 across Spain, Brazil and Germany, with full L4 targeted for 2030.
- T-Mobile: agents inside the IMS of the 5G standalone core, first service live is translation in 50+ languages.
- Deutsche Telekom: RAN Guardian ran 100+ autonomous remediations in its first month on the German network, cutting incident-handling from hours to about a minute.
See you next month!
