Texas data centers flunk ERCOT voltage tests, threatening grid stability

When a Data Center Goes Dark, the Whole Grid Can Follow

ERCOT, the Texas grid operator, has published findings that a number of newly connected data centers and cryptocurrency mining facilities are failing voltage-stability tests — a prerequisite for reliable grid participation. The Reuters report describes ERCOT flagging these sites as risks, though it ran into the Reuters paywall and full specifics of which sites and how many were unavailable. The HN thread filled in the physics.

What a Failed Voltage Test Actually Means

Voltage compliance testing for large industrial loads verifies that a facility can ramp down consumption in a controlled, graduated way rather than cutting power instantaneously. For a residential customer, abrupt disconnection is trivial — the grid barely notices. For a 100-megawatt data center or a 500-megawatt crypto mining campus, a sudden load drop is the electrical equivalent of a mechanical shock to every generator on the interconnect.

ERCOT requires these tests precisely because Texas runs an isolated grid — it has minimal tie-lines to neighboring systems that could absorb disturbances. Large new loads connecting without passing voltage compliance are, in effect, unverified actors in a system with no margin for uncontrolled surprises.

The Thread's Grid Engineers Explain the Physics

In the thread, a commenter with clear electrical engineering knowledge posted the most-cited explanation of why this matters:

"The loads are slowing down the generators that are burning a well metered amount of fuel to stay at 60Hz. This is a delicate balance since the phase angle must also be spot on. If a generator and the local line disagree on f, phase or V, you have a short circuit… Rotating machines are big and heavy, so the first line of defense is their inertia. But this is a finite (and precious) resource. Contrary to belief, renewables, or generally speaking DC, makes this stability problem worse. They generate large amounts of power while providing no inertia… a large enough loss of load causes the grid to enter into unstable oscillations, causing protective relays to trip causing a zipper effect where the grid goes down. Now restart will take a few days depending on the energy mix." — a power systems engineer in the thread

This is the reversal the headline doesn't capture: the problem isn't that data centers consume too much power. It's that their load profile — DC-heavy, with no rotational inertia, and capable of dropping to zero instantly — makes the grid more fragile in the same way an overabundance of solar and wind does. The article frames it as a compliance story; the engineers in the thread frame it as a physics story.

A second reversal came from the same commenter, pivoting to economics:

"How about an across the board $1/W hook up fee for new customers? That's about the price of installed capacity per watt. New house with 200A panel at 30% utilization rate? $3,600. New data center with 80% utilization rates at 100MW? $80 million dollars. New 10 GW data center? That'll be $8 billion. It's outrageous that I'm paying an extra fee to export energy to a neighboring state to power a datacenter." — the same power systems engineer

The current ERCOT connection fee structure doesn't scale this way. Large industrial consumers benefit from the same grid infrastructure that residential ratepayers fund, while imposing stability risks that residential users don't. The voltage test failures are a symptom of that misalignment: sites were connected, or are seeking connection, without demonstrating they can behave safely under grid stress.

A third voice asked the question that puts batteries at the center:

"I'm curious how this works for other large industrial consumers. Do they have some kind of artificial load that lets them gradually reduce consumption instead of doing it all at once?" — an HN commenter

The answer, implicit in the thread, is yes — aluminum smelters, for instance, can ride through short disturbances using the thermal mass of their process. Data centers and crypto mines have no equivalent natural buffer unless they install large battery systems specifically for load-smoothing, which most have not.

The Structural Bet ERCOT Is Making

Voltage compliance testing is the right tool. The question is enforcement timing: ERCOT is flagging risks on sites that are apparently already operating or in late-stage connection approval, which means the corrective action is retrofitting rather than pre-qualification. The power-systems engineer in the thread framed the political economy cleanly: large new loads are being subsidized by existing ratepayers both in infrastructure cost and in grid-stability risk. Failing a voltage test should mean you don't connect — not that you get a grace period to install the hardware you should have planned for in the first place.