UBS Has Identified Power Grid Limits Hindering AI

Power supply bottlenecks may stall the rapid growth of the AI industry as infrastructure development trails construction.

Updated on Oct. 12, 2026 in Data Centers

UBS Has Identified Power Grid Limits Hindering AI

Live Poll

Do you expect your monthly electricity bill to rise due to increased regional data center construction?

UBS analysts have identified the power grid as the primary constraint on the expansion of the AI industry. The firm initiated bullish coverage on colocation data center stocks, citing a significant gap between electricity demand and available infrastructure.

Why it matters

The discrepancy between data center construction speeds and energy project timelines creates major operational risks for tech firms. This imbalance threatens to delay large-scale AI deployment as power supply fails to keep pace with demand.

Data centers require baseload power comparable to a mid-sized city, yet large transformer equipment lead times have surged to 210 weeks. Additionally, the U.S. interconnection queue currently holds 2.3 terawatts of capacity awaiting grid access.

The players

UBS

This global financial services firm provides investment banking, asset management, and wealth management services to institutional and private clients.

The details

Infrastructure development requires 5 to 10 years to complete, significantly outlasting the 12 to 24 months needed to construct a new data center. Regulatory hurdles and community opposition, such as a recent one-year moratorium on large-scale data centers in New York, further complicate the expansion of reliable energy supplies.

2024 served as the baseline for U.S. installed power capacity.
September 23, 2026, marked the date UBS initiated coverage on colocation stocks.
2027 is the target year for projected data center power demand and the onset of expected growth.
2027-2029 represents the forecasted period for 80 percent annual EBITDA growth for colocation firms.
2030 serves as the benchmark year for meeting a 415 gigawatt power availability goal.

The Tech Race

This focus on grid capacity highlights a shift in the data center arms race from purely architectural capacity to energy security and access. This trajectory marks a departure from historical tech growth periods where hardware and software scaling were rarely constrained by fundamental utility availability.

The power supply bottleneck may result in higher costs for cloud computing services as energy constraints force providers to compete for limited capacity. These infrastructure delays could also affect the rollout speed of new consumer-facing AI applications as developers wait for server availability.

The takeaway

Reliable power access has become the single most important factor in the success of future AI infrastructure projects. Investors and consumers should monitor energy grid expansion progress as closely as they track advancements in artificial intelligence models.