ai
Goldman Sachs Report Identifies Physical Constraints on Agentic AI Expansion
A May 13 Goldman Sachs report states that agentic AI development faces limits from data center capacity, power supply, and workforce shortages rather than memory or chips. The bank estimates the U.S. would need 72 gigawatts of new power generation and hundreds of thousands of additional grid workers through 2030.
May 26, 9:46 AM·1m read1 source
BenzingaAudio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Goldman Sachs analysts said in a May 13 report that the main constraints on agentic AI growth have shifted from semiconductors and memory to physical infrastructure. U.S. power output, a shortage of skilled grid workers, restricted land availability, and extended supply-chain lead times for materials such as steel.
AI systems require 60-130 times more energy than current chatbot models, according to a June 2025 study from the Korea Advanced Institute of Science and Technology that Goldman Sachs cited. U.S. power generation, equal to the output of 72 large nuclear plants, plus 760,000 grid workers and 207,000 experienced transmission and distribution workers.
Sachs noted that investors have concentrated on data centers and chips while overlooking these infrastructure limits. The report pointed to a ninefold difference in earnings before interest and taxes between chip, memory, and server firms and companies involved in power, components, and data center services.
, Seagate Technology Holdings, and Micron Technology Inc. have recorded triple-digit percentage gains this year. "Many investors are still looking to replicate past successes in data centers, missing the critical chokepoints that will define the next phase of growth," the analysts said.
