# AI Infrastructure Race Shifts From Chip Supply to Power Access Morgan Stanley warns of a critical 32-gigawatt electricity shortfall threatening future AI data center deployment. - **Source:** [Economic Times](https://brifing.si/source/37791137-c35a-4ffe-b6a9-fe103afc9e48) - **Published:** 2026-10-17 - **Topic:** [Business](https://brifing.si/c/business) - **Original article:** - **On Brifing:** https://brifing.si/s/8e4aefeb-3773-486f-8f4e-820f5452f0f2 ## The Emerging Power Bottleneck The rapid expansion of artificial intelligence infrastructure is encountering a significant physical constraint beyond semiconductor availability. Morgan Stanley analysis indicates that access to electricity and grid infrastructure is becoming the primary limiting factor for AI growth. This shift suggests that the industry's trajectory is increasingly dependent on energy availability rather than hardware. ## Projected 2028 Energy Deficit Projections highlight a looming 32-gigawatt power deficit expected to materialize by 2028. This shortfall represents a critical challenge for the scalability of large-scale data centers required for advanced model training. Without sufficient power, the deployment of high-performance chips will be severely hampered, regardless of how many units are successfully manufactured. ## Disproportionate Impact on Enterprises The impending electricity shortage poses a disproportionate risk to smaller enterprises compared to their larger, better-capitalized counterparts. Larger firms may possess the resources to navigate grid limitations, while smaller players struggle to secure necessary power allocations. This disparity could consolidate market power and limit competitive innovation within the broader AI ecosystem. ## Data Center Development Risks The 32-gigawatt deficit directly threatens the development pipeline for new data centers. Developers face significant hurdles in securing grid connections and reliable power sources, which are essential for operational viability. As energy demand surges, the physical limitations of existing electrical grids become a central strategic concern for all major technology stakeholders. ## Persistent Internal Power Dilemmas While some global strategies might offer temporary relief for certain companies, internal power dilemmas within the United States are expected to persist. Domestic grid constraints remain a structural issue that cannot be easily bypassed through international diversification. These localized challenges underscore the necessity of addressing national energy policy alongside technological advancement efforts. ## Shifting Strategic Priorities Industry leaders are forced to re-evaluate their expansion strategies, prioritizing regions with robust energy infrastructure. The focus is moving away from purely chip-centric procurement toward securing long-term power purchase agreements and grid access. This transition marks a fundamental change in how technology companies approach the physical requirements of their digital operations. ## Operational Constraints on Deployment Even with an abundance of chips, the inability to power them renders the hardware ineffective for large-scale AI tasks. The bottleneck is shifting from the supply chain of semiconductors to the capacity of the electrical grid. Companies must now account for energy availability as a core variable in their long-term infrastructure planning. ## Future Industry Implications The findings suggest that the AI race will be won by those who can secure energy, not just silicon. By 2028, the ability to manage power constraints will define market leadership. Stakeholders must monitor grid capacity and energy policy closely, as these factors will dictate the pace of future technological innovation and deployment. ## Sources covering this story - https://economictimes.indiatimes.com/markets/us-stocks/news/ai-race-may-depend-more-on-access-to-power-infrastructure-than-chip-supply-morgan-stanley/articleshow/135012540.cms — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._