Fed official flags AI 'too big to fail' risk; Valley sees its own exposure

Key Takeaways

Kansas City Fed’s Jeff Schmid warned the AI industry could become too big to fail.
San Joaquin County supervisors opened a process Sept. 2 to study AI data center impacts.
PG&E projects 1.4 GW of new data center load online between 2026 and 2030.
Patterson imposed a 45-day pause on data centers in July over power and water concerns.

"Too big to fail" is back in the conversation, this time about artificial intelligence. The warning came from Kansas City Federal Reserve President Jeff Schmid, who said the central bank needs to understand whether the growing AI "ecosystem" is getting so large and interconnected that a failure could spill across the economy.

San Joaquin County’s Board of Supervisors has already started asking its own version of that question. On Sept. 2, the county opened a file on how large AI data centers could affect noise, power and land use. The nut graf here is simple, if the money and infrastructure concentrate in a few big players and projects, the Valley’s banks, utilities and counties carry the downside when something snaps.

What the Fed worry means

Schmid’s point was about interconnections, not hype. If a handful of companies and lenders dominate the chips, cloud contracts and data center build‑outs, then a break in one part can hit the rest. That is a financial‑stability question the Fed tracks in other sectors. It now has to learn the plumbing of AI too.

Local lenders are exposed through construction loans and equipment financing. Regional banks do not disclose every tech customer, but credit officers here have said for months that data center proposals come with tight timelines and hard power asks. Worth watching.

Where the Valley fits

Power is the first pinch point. Pacific Gas and Electric says it has about 1.4 gigawatts of new data center load in final design to come online between 2026 and 2030, part of a larger queue across Northern and Central California. Counties are reacting. Patterson’s council in Stanislaus County approved a 45‑day pause in July to study water, energy and noise before taking more applications. San Joaquin County is doing similar due‑diligence work now.

UC Merced and ag processors are smaller in comparison, but they’re part of the same web. When cloud costs spike or access tightens, research labs and plant-floor AI pilots can stall. That hits schedules, then payrolls.

Who could feel it first

If credit tightens, community banks and municipal utilities feel it early. A delayed substation or interconnection can strand a half‑built campus, which strands the loans behind it. County planners get heat next, because entitlement clocks keep ticking even when gear is on backorder. And if a major AI vendor changes terms, local firms that built their workflows around one tool can eat real switching costs.

Schmid didn’t say the sky is falling. He said the system needs a map before it bets the farm on a single road. On the clerk’s desk at the supervisors’ chambers in Stockton, a green binder with dog‑eared tabs sat next to the mic.

Central Valley AI is produced by the CVAI Newsdesk team and developed by Kaweah Tech, a regional firm that builds, deploys, and integrates AI solutions for businesses across California's Central Valley.

Source

https://www.mercedsunstar.com/news/business/article317410896.html