The historical claim in the lede is well documented. On 1 and 2 September 1859, a coronal mass ejection from the Sun struck Earth’s magnetosphere with such force that telegraph systems across North America, Europe, and parts of Asia and Australia were disrupted. Operators in some stations received electric shocks. Paper caught fire from sparks leaping from telegraph keys. A few stations burned. On the night of 2 September, two operators on the Boston-to-Portland line held a conversation over wires they had disconnected from their batteries, exchanging messages on what they called the auroral current alone.
The forward claim in the lede is more complicated than it sounds. It is worth slowing down on exactly what is being said, by whom, and on what basis.
What “scientists warn” is actually doing
The phrase “scientists warn” tends to flatten a set of distinct claims into one chorus. In the case of severe space weather, that chorus is largely composed of three documents that get cited again and again in the popular coverage.
The first is the 2008 National Academies workshop report, Severe Space Weather Events: Understanding Societal and Economic Impacts. The second is a 2013 study by Lloyd’s of London and Atmospheric and Environmental Research, which produced the often-quoted estimate that a Carrington-class event today could cost the United States economy between $0.6 trillion and $2.6 trillion. The third is a Lloyd’s update in 2024 that revised global economic losses to a range of $1.2 trillion to $9.1 trillion over a five-year recovery period, depending on the severity scenario.
None of these are scientific consensus documents in the textbook sense. They are scenario models built by insurance researchers, risk consultancies, and committees of researchers asked what an extreme event could do to specific infrastructure. The headline numbers in popular coverage are the high ends of these scenario ranges, not the midpoints.
The peer-reviewed literature on geomagnetically induced currents and transformer vulnerability is real and growing. What it does not produce is a single confident statement of the form “the next Carrington event will knock out power grids across entire continents.” It produces a range of estimates with wide error bars, contingent on the storm’s direction, intensity, duration, and Earth’s orientation at the moment of impact.
The Quebec data point
The closest real-world test we have is the March 1989 storm that brought down Hydro-Québec’s transmission system. The storm was substantially weaker than the 1859 event, by most reconstructions. Within 90 seconds, six million people lost power. The full restoration took about nine hours, with parts of the system out for longer. The cause was a voltage collapse triggered by geomagnetically induced currents in long high-voltage lines crossing a region of resistive bedrock.
The 1989 event is useful because it is not a model. It happened. A storm of moderate severity, by Carrington standards, took down a major utility for nearly a day. The inference from there to a Carrington-class event is not a guarantee, but it is not pure speculation either.
The transformer problem
The specific concern in most modern grid studies is not the lights going off for an afternoon. It is the loss of extra-high-voltage transformers. These units are large, custom-built, and have lead times measured in months to years. The North American grid maintains spare inventory sufficient to cover less than ten percent of the installed base, according to industry estimates summarised in the Lloyd’s work and in subsequent threat-analysis reports.
The scenario that sits behind the more dramatic loss estimates is one where a severe storm damages or destroys a meaningful fraction of these transformers across multiple utilities simultaneously, and replacement is bottlenecked by global manufacturing capacity. In that scenario, the affected population is offline not for hours, but for weeks to months. A 2013 Lloyd’s projection suggested 20 to 40 million Americans could face outages ranging from 16 days to one or two years, with duration dictated by transformer availability.
This is the scenario the popular phrase “knock out power grids across entire continents” is reaching for. The phrase compresses a contingent, modelled scenario into a confident prediction. Worth keeping that gap visible.
What we cannot estimate well
The frequency of Carrington-class events is genuinely uncertain. Estimates of the probability of a similar event in any given decade range from roughly one percent to ten percent in the published literature, with the upper end driven by aurorae records and ice-core reconstructions that themselves are contested. The 2012 coronal mass ejection that missed Earth by about a week of orbital geometry is often cited as evidence that the underlying solar activity has not become less extreme since 1859.
The other thing that is hard to estimate is forecast time. A coronal mass ejection takes roughly 15 to 18 hours to cross from the Sun to Earth in the most severe cases, and longer in moderate ones. That window is the operational margin utilities have to take protective action. Whether grid operators would, in fact, take that action, and whether the action would be sufficient, are questions that depend on coordination, regulation, and the specific configuration of the storm. None of these are settled.
What the lede gets right and what it overstates
The historical half of the lede is accurate and well sourced. The 1859 storm did hit Earth hard. Telegraph operators were shocked. Sparks did jump from equipment. Some stations did burn.
The forward half is a compressed version of a more careful claim. A severe geomagnetic storm today could cause widespread outages, large economic losses, and prolonged disruption in worst-case scenarios. Whether it would, and at what scale, depends on parameters that no one currently models with the kind of confidence the phrase “scientists warn” implies.
The next milestones to watch are technical: continued upgrades to the US Space Weather Prediction Center’s forecasting capacity, the ongoing rollout of the European Space Agency’s Vigil mission to monitor the Sun from the L5 Lagrange point, and the slow progress of grid hardening regulation in jurisdictions exposed to high-latitude induced currents. The story is not whether the Sun is dangerous. It is.
The story is how much of the danger we have actually built ourselves into.
