AI Boom is Outgrowing Digital Infrastructure

An overwhelming majority of technology and business leaders in the US and Europe believe that current networks will require substantial evolution and investment to meet the demands of the AI supercycle, according to research commissioned by Nokia. The research surveyed around 2,000 technology and business decision-makers in the US and Europe, including telecommunication and data center infrastructure providers, as well as businesses and organizations planning to adopt and integrate AI into their operations.

The findings highlight a shared recognition across the connectivity ecosystem that next-generation network capabilities are essential and must modernize to support increasingly complex AI workloads. This presents an opportunity for collective action across industry and government to strengthen the digital foundation on which future AI innovation depends.

The research explains why this evolution matters, with AI redefining network requirements. Workloads are becoming more uplink-heavy, data flows are more distributed, and expectations around latency, throughput, resilience, security, and energy efficiency are rising. These changes carry implications not only for telecommunication providers, AI and cloud providers, and mission-critical enterprises, but for national competitiveness and long-term digital leadership.

AI applications – from autonomous vehicles and smart manufacturing lines to surveillance drones and remote health care diagnostics – generate large volumes of data at the edge that must be transmitted upstream for processing, making them uplink-intensive. This stresses today’s networks, which were originally engineered for downlink-focused consumer use, such as browsing websites and video streaming.

While the US continues to lead global AI deployment and mass-market adoption, 88% of US respondents expressed concern that the expansion of network infrastructure may not keep pace with AI investment. Respondents identified bi-directional data flow optimization, expanded fiber capacity, real-time training feedback, and low-latency edge infrastructure as essential priorities and building blocks for modernizing network architecture and powering the next phase of AI growth.

In Europe, 86% of enterprise respondents said current networks are not yet equipped to handle widespread AI adoption. Two-thirds of those surveyed said they already have AI in live use, and more than half have already experienced challenges such as downtime, latency, and throughput constraints associated with increasing data demands. To address these challenges, respondents emphasized the need for consistent regulatory simplification and alignment across markets, timely spectrum availability, adjustments in competition policy to enable market consolidation, and industry-wide investment in energy-efficient, AI-ready networks.