Power shortages and execution gaps delay $64 billion in data centers

The shift from demand-driven growth to execution-based constraints signals a maturing but strained infrastructure market. For streaming and AI firms, these AI data center bottlenecks mean that even fully funded projects face multi-year delays if they cannot secure local social licenses or grid priority. This creates a competitive environment where the ability to navigate local energy regulations becomes as critical as the underlying technology stack. As the supply chain grows more complex, the industry must move away from fragmented tracking systems to avoid further schedule slippage. Watch for whether developers begin prioritizing regions with less-taxed power grids over traditional high-density hubs to bypass these delivery hurdles.

The power bottleneck identified by Salesforce and Accenture executives reflects a broader infrastructure crisis affecting cloud and streaming workloads alike. In June 2026, Nokia announced partnerships with AWS and Databricks to build a unified data and control layer for autonomous network operations, positioning its Autonomous Network Fabric as a cloud-hosted orchestration system that integrates OSS applications, AI services, and intent-based automation. The move underscores how telecom operators, facing similar compute and energy constraints, are consolidating fragmented data environments into unified platforms to reduce operational overhead and accelerate AI deployment timelines.

The competitive dynamics around AI infrastructure investment are intensifying as vendors differentiate their approaches to compute architecture. Ericsson and Nokia are diverging sharply on AI-RAN strategy, with Nokia building its entire RAN roadmap on Nvidia GPU-accelerated compute following Nvidia's $1 billion investment in the Finnish company, while Ericsson anchors its strategy in standalone 5G cores and programmable RAN. This architectural split has direct implications for data center demand: GPU-heavy AI-RAN deployments require significantly more power density per rack than traditional baseband processing, compounding the grid constraints that the Salesforce and Accenture panel identified.

Ericsson's own infrastructure roadmap highlights the scale of the challenge. The company projects uplink traffic could triple over the next five years, driven by AI glasses, persistent voice interaction, sensors, and real-time video, with uplink growth already outpacing downlink growth by 50% in roughly a third of operator networks. Ericsson describes the network as becoming an "intelligent fabric" that must host AI inference at the edge rather than relying solely on centralized data centers, a distributed approach that could alleviate some pressure on grid-constrained hyperscale facilities. Meanwhile, Verizon disclosed that its 60,000-site vRAN network is now applying agentic AI to configuration changes and network optimization, while publicly calling for industry-wide interoperability standards for agentic systems. That standards gap mirrors the fragmented data systems the Salesforce panel flagged as a root cause of project delays, suggesting that both the streaming and telecom sectors face similar coordination failures as they scale AI workloads against finite power resources.