Could Space Weather Risks Cost the U.S. Economy $1.81 Billion a Day?

Key Takeaways

• A 2026 study estimates $1.81 billion in daily U.S. losses under a severe geomagnetic storm scenario.
• Local geology and grid design shape exposure, so storm intensity alone cannot predict damage.
• The estimates exclude outage duration and some failure mechanisms, leaving total costs uncertain.

A New Estimate Connects Space Weather Risks to Economic Losses

A study published on September 4, 2026, estimates that a severe geomagnetic storm could disrupt electricity for approximately 5.1 million people and 135,000 businesses in the United States, producing economic losses of about $1.81 billion per day. The research published in AGU Advances, led by Edward J. Oughton and Dennies K. Bor, connects space weather risks to the equipment that carries electricity and the economic activity that depends on it.

The estimate describes a modeled storm scenario with a 250-year return period. It does not predict an approaching event, establish a blackout duration, or calculate the complete cost of a national disaster. Its subject is narrower: potential daily economic disruption associated with transformer heating under specified assumptions about the storm and the power network.

That distinction matters because a dollar figure can travel much farther than the conditions attached to it. A daily loss estimate cannot be treated as a total damage bill. Nor does the modeled population represent people who have already experienced an outage or are certain to lose power during the next strong storm.

For the less severe 100-year scenario, the researchers estimate that approximately 3.45 million people and 91,000 businesses could experience electricity disruption during the modeled outage day. Estimated total economic losses reach $1.22 billion per day. The progression between scenarios indicates greater exposure as storm severity increases, although local outcomes also depend on equipment characteristics and network connections.

The scientific contribution lies in how the researchers connect these outcomes. Instead of starting with an assumed blackout covering a large region, they connect estimated electrical conditions at the ground to transmission infrastructure, then translate possible service interruptions into economic consequences.

In its September 4, 2026, commentary on the findings, Eos, published by the American Geophysical Union, emphasizes the value of combining disciplines that often assess the problem separately. It also draws attention to missing interactions, including the possibility of a geomagnetic storm occurring during another emergency.

The result offers a more explicit basis for examining infrastructure exposure. It also makes assumptions easier to inspect, including where public grid information is incomplete and where the economic model simplifies business behavior.

How Solar Activity Produces Electrical Stress on Earth

A coronal mass ejection is an eruption of plasma, an electrically charged gas, carrying a magnetic field away from the Sun. When an Earth-directed eruption reaches the planet, its interaction with Earth’s surrounding magnetic environment can produce a geomagnetic storm. The strength of that interaction depends partly on the orientation and persistence of the arriving magnetic field.

The National Oceanic and Atmospheric Administration explains that geomagnetic storms involve an efficient transfer of energy from the solar wind into Earth’s magnetic environment. Changes in electrical currents above the planet produce magnetic disturbances that extend to the ground. Those changing magnetic conditions induce electric fields within the Earth.

Long transmission lines can collect the resulting electrical effects over substantial distances. Currents then flow through grounded portions of the power network, including transformer connections. These are called geomagnetically induced currents, meaning currents created in conducting networks by changes in the surrounding magnetic environment.

Transformers normally operate with alternating current, whose direction reverses repeatedly. The much slower currents associated with geomagnetic disturbances can shift the magnetic operating conditions inside them. This can distort normal operation and create additional heating in parts of the equipment.

The power transmission explanation provided by the federal Space Weather Prediction Center describes the connection between geomagnetic activity and grid disturbances. Equipment stress is only part of the problem. Distorted electrical conditions can also interfere with protective devices and make it harder for operators to maintain stable voltage.

A damaged transformer and a protective disconnection are different outcomes. Equipment may disconnect without suffering permanent damage, and a transformer can experience stress without immediately interrupting electricity service. The Oughton study uses probabilistic relationships to represent malfunction or protection-system misoperation as a proxy for substation failure, rather than predicting that every exposed transformer will be destroyed.

Solar flares should also be distinguished from coronal mass ejections. Flares release electromagnetic radiation, and their effects on radio communications differ from the ground-current pathway examined in the study. The space weather scales used in public communication separate geomagnetic disturbances from other solar hazards.

This separation prevents an error in interpreting solar news. A powerful flare designation alone does not specify the electrical stress at a particular substation, and an impressive auroral display does not provide a measurement of transformer damage.

Building a Model From Ground Conditions to Business Disruption

The researchers assembled a representation of the contiguous United States transmission network containing 10,464 substations and 16,256 transmission lines. The modeled infrastructure operates at voltages of at least 161 kilovolts. It combines substation information from OpenStreetMap with transmission data from the federal Homeland Infrastructure Foundation-Level Data program.

This is a national research model assembled from available information, rather than a complete operating model supplied by every utility. Its scale permits analysis across the country, but the underlying records do not reveal every transformer configuration or protective installation. The researchers address some of these gaps by testing plausible equipment characteristics.

The physical calculation begins with magnetic observations and information about how the ground conducts electricity. These inputs support estimates of surface electric fields during storms. Statistical analysis then produces scenarios associated with return periods of 100 to 250 years.

Transmission lines respond differently depending on their length and orientation. The model estimates induced voltages along those lines and calculates how currents move through the represented network. It then estimates the electrical stress experienced by transformer windings, the conductors inside the equipment.

A further step translates stress into failure probability. The researchers use mathematical vulnerability curves to represent how malfunction becomes more likely as exposure increases. Their calculations also account for age-related deterioration, recognizing that identical electrical exposure need not produce identical outcomes across a mixed equipment fleet.

To examine uncertainty, the model repeatedly samples possible configurations and equipment parameters. This approach, known as Monte Carlo sampling, runs many versions of the calculation with different plausible inputs. The resulting distribution describes how estimates change within those assumptions.

Population and business activity must then be associated with substations. The study constructs approximate geographic service areas around substation locations and assigns economic activity to them. These areas support national analysis, but they are not verified customer-by-customer electricity service territories.

The economic component uses relationships between industries to estimate effects beyond the immediately affected area. A reduction in activity can lower demand for inputs supplied elsewhere, extending losses beyond businesses whose electricity stops.

Each connection introduces assumptions that need separate scrutiny. A good estimate of the ground electric field does not automatically establish an accurate outage footprint, just as an accurate outage footprint does not determine how quickly businesses recover.

The framework’s value is that these connections are visible. Better utility records can improve the engineering component without requiring the entire economic model to be rebuilt, and updated business data can improve the economic calculation without changing the underlying physics.

Why Geography Changes the Risk

Wisconsin and Minnesota appear among the study’s concentrations of modeled vulnerability. Other exposed areas include parts of the upper Midwest and the northeastern coast. The results also identify elevated geoelectric responses near the Great Lakes and along portions of the Appalachian region.

These patterns cannot be explained by latitude alone. The ground beneath a transmission corridor changes how magnetic disturbances translate into electric fields. The network above it determines how those fields drive currents through equipment.

The relevance of local geology predates the new economic framework. A 2020 geoelectric hazard study, documented by the U.S. Geological Survey, estimated once-per-century electric field strengths that differed by more than three orders of magnitude among surveyed locations. Its findings demonstrate why a national storm classification cannot substitute for local exposure analysis.

The Oughton study similarly combines geographic hazard with network characteristics. Line direction affects the induced voltage along a transmission path, and grounding conditions influence the routes available to current. Transformer design then affects how that current translates into stress.

Economic exposure introduces another layer. An electrically vulnerable location serving substantial business activity can generate larger estimated losses than a location with similar physical exposure but less concentrated production. Physical hazard and economic consequence are connected, but they are not interchangeable measures.

The study’s geographic limitations deserve equal attention. Its network illustration explicitly identifies sparse coverage in Washington and New York because the OpenStreetMap query returned limited substation information for those states in the 2024 data used. A sparsely represented area cannot be interpreted as a reliably low-risk area.

Nor should a modeled concentration of vulnerability be read as an operating instruction for a named utility. Actual exposure depends on information that the national model does not fully possess, including equipment condition and protective measures already installed.

For planning purposes, the maps can help identify places where additional measurements and utility participation would be valuable. They cannot establish that all substations within a named state share the same vulnerability, or that infrastructure outside the strongest modeled concentrations requires no preparation.

The appropriate geographic unit for an investment decision may be a specific equipment group or transmission corridor. Statewide labels can communicate a research pattern, but they conceal differences that determine which intervention would reduce losses.

What the Daily Economic Figures Actually Measure

Under the 250-year scenario, the study estimates approximately $980 million in direct economic losses per day. Once inter-industry effects are included, the total reaches about $1.81 billion. The roughly $830 million difference is calculated from the rounded published figures and represents modeled economic effects beyond the direct loss component.

The 100-year scenario produces approximately $660 million in direct daily losses and $1.22 billion in total daily losses. Intermediate scenarios produce total estimates of $1.46 billion for a 150-year event and $1.65 billion for a 200-year event. These are conditional outputs from the 2026 study, not observed losses from historical storms.

The economic mechanism uses an input-output model, which describes how industries purchase from and sell to one another. The Bureau of Economic Analysis explains these relationships through its input-output accounts. The researchers estimate a disruption associated with the affected population and expenditure, then propagate it through those economic relationships to capture potential reductions in activity outside the modeled outage area.

Such calculations are useful for estimating short-term interdependence. They do not track every contract or shipment, and they cannot fully represent how individual businesses change behavior during a disruption. The study acknowledges that productivity differences within industries are simplified.

Its data also have identifiable reference years. The economic structure draws on 2023 supply-use tables from the Bureau of Economic Analysis, and population information comes from the 2020 census. Publication in 2026 does not mean every underlying observation describes the economy in 2026.

The sector results place manufacturing among the larger modeled contributors to losses. Finance and real estate also feature prominently, alongside a combined education and entertainment grouping. These results reflect the economic composition of affected locations and the model’s allocation method, rather than a universal ranking of industries most vulnerable to every solar storm.

Backup electricity is another unresolved issue. The researchers do not explicitly model backup arrangements at facilities such as data centers and hospitals. Backup systems could preserve some activity, although operating them can introduce additional costs. Their presence complicates any direct translation from a nearby substation failure to a complete loss of local production.

Most importantly, the study deliberately avoids selecting an outage duration. It reports daily values because restoration depends on the actual equipment affected and the options available to operators.

Multiplying $1.81 billion by an assumed number of days would produce an arithmetic scenario, not a finding established by the research. A defensible event-total estimate would need to account for changing outage boundaries and the pace of restoration. It would also need to distinguish activity permanently lost from work delayed and completed later.

The Uncertainties Behind the Headline Number

The study reports a 95% confidence interval of $1.65 billion to $1.96 billion for total daily losses under its 250-year scenario. That interval reflects uncertainty in sampled model inputs. It does not capture every uncertainty in storm behavior, infrastructure representation, or economic response.

The distinction is important because the interval looks relatively narrow beside the scale of the potential disruption. It cannot be interpreted as establishing that an actual future storm has a 95% probability of producing losses inside those bounds. The result remains conditional on the model’s structure and scenario construction.

A 250-year return period also requires careful interpretation. Under a stationary annual exceedance interpretation, it corresponds to an annual exceedance probability of approximately 0.4% for the specified hazard level. It does not mean that such an event occurs on a fixed schedule or that a previous event creates a protected interval.

There is a further complication in this research. The return-period scenarios use statistical peak electric-field amplitudes estimated separately at individual locations. The modeled network is exposed to those local peak amplitudes simultaneously, rather than to a single observed storm snapshot.

That construction is useful for testing the network under demanding conditions, but it does not establish the joint probability of all those peaks occurring together. The national economic outcome should not automatically be assigned the same annual probability as an individual site’s return-period hazard.

The authors characterize the losses as generally providing a lower bound because they exclude voltage collapse and cascading grid failures. They also omit restoration costs and explicit disruption across dependent infrastructure systems. Those exclusions could leave substantial consequences outside the calculation.

Other assumptions can push estimates in the opposite direction. Simultaneous site-level peaks can create a demanding exposure scenario, and the framework does not explicitly represent operational network changes intended to reduce storm-related currents. Backup electricity could also reduce business losses relative to a simplified outage calculation.

Consequently, the lower-bound description needs to remain attached to the omitted mechanisms. It is not a demonstrated guarantee that every real event assigned a comparable severity would cost more than $1.81 billion per day.

Extreme-event statistics introduce additional uncertainty. Long return periods must be inferred from a limited observation record, and the study assumes a particular statistical distribution for extreme electric-field values. Additional measurements or another defensible distribution could change the estimated tail of the hazard.

These limitations do not erase the research contribution. They define the difference between a national screening framework and a complete operational forecast. The former can identify where better information matters; the latter would need much more detailed knowledge of the storm, the equipment, and the grid’s condition when the event occurs.

What the May 2024 Storm Reveals About Validation and Protection

During the May 2024 Gannon storm, North America’s bulk power system remained stable and continued serving all connected load throughout the three-day event. An April 30, 2026, reliability presentation by the North American Electric Reliability Corporation reports that early notification helped operators prepare and implement mitigation measures. It also records limited equipment outage and degradation reports.

This observed outcome places the new loss estimates in perspective. A strong geomagnetic event does not automatically produce the outage footprint calculated for a different statistical scenario. Grid condition and operator response affect what happens after the physical disturbance arrives.

The Oughton team used measurements from the Tennessee Valley Authority during the 2024 storm to evaluate its current calculations. Comparing modeled currents with measured currents provides a test of the physical and engineering links. It does not directly validate the hypothetical national economic losses.

Performance differed substantially among monitoring locations. The paper reports correlation magnitudes ranging from 0.02 to 0.70, alongside prediction-efficiency values extending below zero. The authors find stronger agreement in slower variations relevant to transformer heating, with short-period peaks underestimated.

These results support a qualified interpretation of validation. The framework captures useful aspects of the electrical response, but it does not reproduce every measured time series with equal accuracy. Describing the complete economic estimate as proven by the 2024 event would exceed the evidence.

New Zealand provides a separate example of protection in practice. A 2025 study of its response documents how Transpower used planned transmission-line disconnections during the Gannon storm to reduce currents at vulnerable transformers. Electricity supply to customers continued without interruption.

At a transformer in Dunedin, measured current reached approximately 113 amperes after mitigation. Modeling suggested a peak near 200 amperes could have occurred without the protective network changes. The larger value is a calculation of what might have happened without mitigation, not a second measurement.

The New Zealand experience demonstrates that network operation can change exposure. It also shows why protection cannot be reduced to disconnecting equipment indiscriminately: changes in one location can redistribute currents elsewhere.

Taken together, the operational evidence supports preparation without implying immunity. Successful performance during one event does not establish how the same network would respond to a stronger disturbance or a different spatial pattern. It does show that operators influence the outcome, a factor that national loss models need to represent more fully.

Turning Risk Estimates Into Investment Decisions

The research identifies several possible uses for coupled modeling, including selecting monitoring locations and evaluating protective investments. Its economic outputs could help compare the consequences of equipment failures that appear similar in engineering terms. A substation’s importance depends partly on the services and economic activity connected to it.

A practical investment process would begin by improving information where uncertainty could change the decision. The paper identifies missing transformer characteristics and grounding values as constraints on local accuracy. Access to utility records could replace broad assumptions with equipment-specific inputs.

This does not require publishing sensitive infrastructure details. The analytical need is for qualified researchers and operators to work with accurate information under appropriate arrangements. A public national model and a confidential utility assessment serve different purposes.

Monitoring has value beyond supplying a larger dataset. Measurements collected during storms can test whether modeled currents appear where expected and whether planned operational changes reduce exposure. The Tennessee Valley Authority comparison shows how observations can reveal both useful agreement and local weaknesses.

Protective investments should then be assessed against a defined comparison scenario. The relevant calculation compares expected outcomes with and without an intervention, including the cost of maintaining it. The study does not establish a universal return on investment for a particular device or operating procedure.

Daily economic loss alone is insufficient for that decision. Expected benefits also depend on event probability and the reduction in outage duration. A measure that protects equipment but does little to preserve service can have a different economic effect from one that allows service to resume quickly.

Distribution matters as well. A ranking based entirely on measured economic production would not fully represent public-service obligations or the consequences for people dependent on continuous electricity. The study’s population estimates help broaden the assessment, but they do not monetize every social consequence.

Financial users face a similar boundary. A modeled reduction in economic activity is not automatically an insured claim or a utility’s cash loss. Any application to insurance would need separate analysis of the covered assets and the terms defining a payable loss.

For regulators and planners, the framework offers a way to make assumptions inspectable. Proposed spending can be connected to a specific failure mechanism and a measurable reduction in exposure. Where the result changes sharply with uncertain inputs, acquiring better data may be more valuable than treating the central estimate as settled.

Why the Implications Extend Beyond the U.S. Power Grid

The framework was designed so that other countries could substitute their own geological information and electricity networks. National population and economic data would also need to replace the U.S. inputs. Transferring the method does not mean transferring the dollar estimate.

A country with different transmission architecture can experience different equipment stress under comparable magnetic conditions. Its economic structure also changes the consequences of any outage. The New Zealand experience demonstrates the value of adapting analysis to the actual network and its operating procedures.

Cross-border losses remain another extension rather than a quantified result of the present study. The authors discuss using multi-regional economic models to examine international supply relationships, but their reported U.S. total does not constitute a global loss estimate. Applying a simple multiplier to $1.81 billion would not resolve that gap.

Space weather risks also extend beyond terrestrial transformers. The National Aeronautics and Space Administration describes how changes in the upper atmosphere alter the drag experienced by satellites. Heating and subsequent cooling can change orbital behavior, complicating predictions of spacecraft and debris motion.

Coverage of solar storms and satellite reentries provides related space-industry context. These orbital effects are distinct from transformer heating and are not included in the study’s daily power-disruption estimate. Combining them would require separate modeling and attention to overlapping economic consequences.

The commercial relevance also reaches users of satellite services. As described in coverage of space-based applications across industries, economic value often comes from combining space-derived information with terrestrial systems and business processes. Assessing disruption requires attention to that complete service chain.

This creates a potential use for forecasts that translate space conditions into operational consequences. A utility needs information about likely electrical stress, rather than an aurora forecast alone. A satellite operator needs estimates relevant to orbital conditions and spacecraft operations.

The official geomagnetic storm scale communicates possible effects across severity categories. It is valuable for public awareness, but its categories do not specify a particular company’s loss or the condition of an individual asset. Commercial risk products would need to connect those broad warnings to customer-specific exposure and decisions.

The Oughton framework suggests how that connection could be developed for electricity-dependent activity. It does not establish a market size for such products or prove that a particular forecast service will prevent losses. Demonstrating value would require evidence that information changes decisions and improves outcomes.

Summary

The September 2026 study gives a more explicit account of how a severe geomagnetic disturbance could become an economic disruption. Its estimate of $1.81 billion in daily U.S. losses applies to a specified scenario involving transformer-related service failures. It leaves outage duration and several additional damage mechanisms unresolved.

The most useful advance may be the ability to identify which missing information changes a decision. If uncertainty about transformer configuration substantially alters local exposure, better equipment data has measurable analytical value. If protective switching changes the modeled outcome, operator behavior belongs inside the assessment.

Future comparisons should also distinguish changes in the hazard from changes in preparedness. A lower modeled loss after improved data collection could reflect corrected assumptions rather than physical upgrades. A lower loss after tested protection measures would represent a different kind of progress.

For public planning, the objective is a documented connection between expenditure and service continuity. National loss estimates can establish scale, but investment decisions require evidence about particular equipment and feasible responses. The next improvement in space weather risk assessment will depend as much on access to operating data and observed protective performance as on a more precise national dollar figure.