89% of Telecoms Plan to Increase AI Budgets as Network Automation Overtakes Customer Service as Top Priority

By Hector Herrera | April 28, 2026 | Telecom

Eighty-nine percent of telecom companies plan to grow AI spending in the next 12 months — up from 65% in 2025 — and for the first time, network automation has surpassed customer experience as the industry's top AI use case. Carriers have concluded that AI-operated networks aren't a long-term aspiration. They're a near-term competitive requirement.

That's the central finding from NVIDIA's 2026 State of AI in Telecommunications survey, which captures how decisively the telecom industry's AI investment thesis has shifted in a single year. The jump from 65% to 89% planning budget increases isn't incremental — it represents a near-consensus bet that AI infrastructure is now core to staying competitive in the carrier market.

The Shift to Network Automation

For the past several years, telecoms' flagship AI use case was customer service: AI chatbots, predictive churn models, sentiment analysis for call centers. Those applications are still live, but they're no longer driving investment decisions.

Network automation is now the priority. Carriers are investing in AI systems designed to:

• Self-configure network parameters dynamically as demand patterns shift
• Self-optimize traffic routing in real time, reducing latency and improving throughput without human intervention
• Self-recover from failures autonomously, detecting faults and rerouting traffic before a service desk ticket is even opened

The industry term for this architecture is self-healing networks — infrastructure that continuously monitors itself and corrects problems faster than any human operations team could. The appeal is straightforward: as 5G networks grow in complexity and enterprise customers demand service-level agreements (SLAs) carriers can't reliably deliver with manual operations, automation becomes the only viable path.

Why the Numbers Jumped

The 24-point jump in planned AI budget increases between 2025 and 2026 reflects a few converging pressures:

Competitive necessity. Early movers on AI-driven network automation are reporting measurable gains in operational efficiency and SLA compliance. Carriers that aren't investing are watching rivals claim "AI-powered reliability" in enterprise sales pitches — a message that resonates with IT buyers who've been burned by outages.

5G complexity. Fully deployed 5G networks — with network slicing, edge compute, and dynamic spectrum allocation — are too complex to manage effectively with human-operated playbooks. AI isn't optional for these architectures. It's the operating system.

Opex pressure. Telecom margins have been compressed for years by price competition and infrastructure investment cycles. AI that reduces the number of human hands required to keep a network running addresses the cost problem in a way capital investment alone cannot.

Where the Investment Is Going

Beyond network automation, the NVIDIA survey identifies several other growth areas in telecom AI spending:

• AI RAN (Radio Access Network): AI applied to the radio layer to optimize spectrum use and signal quality dynamically
• Fraud detection: Real-time identification of SIM swapping, account takeover, and billing fraud
• Predictive maintenance: AI monitoring physical infrastructure — towers, cables, switches — to flag failures before they cause outages
• Enterprise AI connectivity: Carriers positioning themselves as AI infrastructure providers, not just pipe

That last category is the most strategically interesting. A number of major carriers are actively rebranding as AI connectivity companies — arguing that as enterprises run more AI workloads at the edge, the network between the data center and the endpoint becomes a critical performance variable. Whether customers buy that positioning is still an open question.

What to Watch

Watch for the first major carrier to announce a specific SLA tied to AI-driven recovery times — something like "network self-recovery in under 60 seconds." That benchmark, if it holds up in practice, would become a competitive forcing function across the industry. Also watch the labor side: as network automation matures, the skills profile of a telecom network engineer shifts significantly toward AI/ML operations and away from traditional configuration work. Union negotiations at major carriers over the next 12–18 months will reflect that tension directly.