AI Resilience Report forElectrical and Electronics Repairers, Powerhouse, Substation, and Relay

Right now, AI is mostly augmenting powerhouse and substation repairers rather than replacing them. The hands-on parts of the job — swapping circuit breakers, bolting in voltage regulators, repairing relay systems — still need a skilled human in the yard. What AI is changing is the data side: spotting problems early and helping technicians decide what to fix.

An IEEE PES working group found that the practical application of AI and machine learning in power system protection and control has started but is still very limited, while having to meet the same strict reliability, security, and accuracy requirements as any other protection system. Utilities are moving fastest on predictive maintenance and inspection: EPRI's transmission program is testing "mechatronic" robots ^[1] that live inside substations and collect camera, LiDAR, and EMF data autonomously, and EPRI is asking how much of that analysis can be automated. POWER Magazine reports that utilities are deploying augmented reality tools so senior technicians can guide less experienced colleagues remotely, and AI-driven predictive maintenance systems help operators catch issues before they escalate, reducing the need for large maintenance crews while improving reliability.

EPRI also notes that AI is being explored for wildfire detection, grid management, predictive maintenance, and cybersecurity across hundreds of utility use cases ^[2] — but warns that messy utility data is the biggest roadblock.

Adoption will likely be steady but cautious. On the "go fast" side, there's a huge worker shortage: Goldman Sachs estimates the power industry may need more than 750,000 new workers by 2030 ^[3], and POWER Magazine reports 96% of utility leaders say AI is a new strategic focus, while 66% say the talent gap is the biggest obstacle to deploying it ^[4]. That gives utilities a strong reason to invest in AI tools that stretch their existing crews.

On the "go slow" side, substations are safety-critical: a bad relay decision can cause outages or injuries, so regulators demand careful testing. Beta Engineering notes the real challenge isn't the AI tool itself but getting teams to adopt it consistently ^[5], and the American Public Power Association points out that utilities are still learning how to layer intelligent, AI-powered workflows on top of human decision-making ^[6]. The good news for you: the human skills that matter most here — physically repairing high-voltage equipment, troubleshooting weird failures, and judging when a machine's recommendation is wrong — are exactly the skills AI is worst at.

This career looks like one where AI becomes a powerful sidekick, not a replacement.