89% of Telecom Operators Are Increasing AI Budgets in 2026, NVIDIA Survey Finds

By Hector Herrera | April 22, 2026 | Telecom

Eighty-nine percent of global telecom operators plan to increase AI spending over the next 12 months, with more than a third forecasting increases above 10%—and network automation has overtaken customer experience as the top ROI use case, according to NVIDIA's 2026 AI in Telco survey. The shift signals that carriers are no longer deploying AI to improve customer interactions; they're deploying it inside the network itself.

The Numbers

NVIDIA's global survey of telecom operators found:

• 89% plan to grow AI spending in the next 12 months
• 35% forecast increases above 10%
• Network automation is now the top ROI use case—surpassing customer experience applications
• Open RAN deployments are growing at a 39% compound annual growth rate

These numbers represent a meaningful change in where telecom AI investment is going. A year ago, the primary AI use cases in telecom were chatbots, churn prediction, and personalized plan recommendations. Those are edge-of-business applications. Network automation is core infrastructure.

What Autonomous Networks Actually Do

An autonomous network handles three functions that traditionally required 24/7 human operations staff:

Self-configuration: The network detects new conditions—a traffic surge, a coverage gap, a new device type—and reconfigures itself to respond without waiting for a human to notice and act.

Self-healing: When a component fails or degrades, the network reroutes traffic and compensates before a human identifies the problem. Mean time to resolution drops from hours to seconds.

Real-time traffic optimization: Dynamic allocation of bandwidth and resources based on actual demand, not planned capacity models. Networks stop being built for peak and start being built for average, with AI filling the gap.

The operational target for most carriers investing in AI is a network that runs itself under normal conditions and surfaces only genuinely novel problems for human attention. That's a different job description for network operations teams than exists today.

Why Open RAN Is the Enabling Layer

Traditional telecom networks are built on proprietary hardware from a small number of vendors—primarily Ericsson, Nokia, and Huawei. Each vendor's software runs only on their hardware, making it difficult to update AI or optimization layers without replacing physical equipment.

Open RAN (Radio Access Network) disaggregates the software from the hardware, allowing operators to run network software from multiple vendors on standard servers. That disaggregation is what makes AI integration tractable at scale—you can update and retrain the AI layer independently of the radio hardware.

The 39% CAGR for Open RAN deployments explains why AI network spending is accelerating now: the infrastructure capable of running AI-managed networks is being built out rapidly, and operators are investing in the AI layer as the hardware layer matures.

Implications

For network equipment vendors: The competitive dynamics are shifting. AI-native network software companies can now compete on Open RAN platforms that didn't exist five years ago. Traditional hardware vendors are adapting, but the barrier to entry for new competitors has dropped.

For enterprise customers: Self-healing, AI-optimized networks should deliver better uptime and faster issue resolution for businesses that depend on carrier infrastructure. The practical benefit will show up slowly as deployments mature—but SLA improvements from autonomous networks are measurable and contractually significant.

For telecom workforces: Network operations centers (NOCs), which monitor and manage network incidents around the clock, are the function most directly affected by autonomous network capabilities. Staffing models built around 24/7 human monitoring will need to be rethought as AI takes over routine incident detection and response.

For security: An AI-managed network is an AI-dependent network. Autonomous systems that self-configure and self-heal are powerful—and they're also attack surfaces. As networks become more software-defined and AI-managed, the security model for those AI systems becomes critical infrastructure protection, not just IT security.

What to Watch

Whether the 89% planning to increase AI spending convert those budgets into measurable network improvements—specifically whether autonomous networks deliver on SLA promises being made to enterprise customers. The 39% Open RAN CAGR also bears watching: if that rate sustains, the structural shift in telecom infrastructure will be largely complete within three years, which would make the AI capabilities described above table stakes rather than competitive differentiation.

Hector Herrera covers AI in telecommunications for NexChron.