Large blackouts are rare, and official investigations of them are unusually detailed. Two stand out: the August 2003 North American blackout and the April 2025 blackout in Spain and Portugal. Neither was caused by a cyberattack according to the reports we cite, and this post does not claim a future grid collapse. It uses them to show how failures cascade and what that means for planning.
14 August 2003
The US-Canada Power System Outage Task Force final report (April 2004) says the outage affected an estimated 50 million people and 61,800 MW of load across eight US states and Ontario. It began a few minutes after 16:00 EDT. Power was not restored for 4 days in some parts of the United States, and parts of Ontario had rolling blackouts for more than a week. US cost estimates ranged from 4 billion to 10 billion dollars.
The task force wrote that "this blackout could have been prevented." It grouped the causes into four. Group 1: the utility and its reliability council failed to assess the system's inadequacies, particularly voltage instability. Group 2: inadequate situational awareness, including no backup monitoring tools after the primary alarm systems failed. Group 3: failure to manage tree growth near transmission lines. Group 4: reliability organisations lacked effective real-time diagnostic support. The report made 46 recommendations.
Note how ordinary these are: monitoring tools that failed silently, untrimmed trees, gaps in communication between control rooms.
28 April 2025
ENTSO-E's page on the Iberian blackout says it began at 12:33 CEST. The expert panel, 49 members chaired by experts from two operators not affected, published a factual report on 3 October 2025 and its final report on 20 March 2026. ENTSO-E calls it the most severe blackout on the European power system in over 20 years.
Its summary says the blackout resulted from "a combination of many interacting factors," including oscillations, gaps in voltage and reactive power control, differences in voltage regulation practices, rapid output reductions and generator disconnections in Spain, and uneven stabilisation capabilities. These led to fast voltage increases and cascading generation disconnections. Recommendations cover operational practices, monitoring, coordination and data exchange, and adaptive regulation.
What is evidence and what is speculation
Evidence: both reports found multiple technical and organisational causes, not a single trigger. Speculation: any claim that a blackout of this size is imminent for your region. Neither source supports it.
What to do now
- Decide how long you can run on your own power: laptops and phones, a small UPS for network gear, and a plan for the office and for any data room.
- Work out your recovery order. Power returns in stages, and mobile networks and data centres on backup power may not last as long as you assume.
- Keep offline copies of runbooks, contact lists and recovery credentials, in print or on a device that does not need the network.
- Ask each critical vendor how long its facilities run on backup power and how it is fuelled.
- Test a restart from cold. Systems that need DNS, directory or a licence server to start can deadlock if all of them are down together.
How this could have been avoided
For the grid operators, the reports point to better monitoring with backups, clearer voltage control practices and better coordination. For a company the parallel is small: do not rely on one monitoring tool, know your start-up dependencies, and rehearse a cold start.
Why it matters
Cascades are about dependency. A grid fault knocks out comms, which knocks out the tools you would use to respond. A written order of dependencies is the cheapest protection you can have.
Quick check
How many people did the 2003 blackout affect, per the task force?
An estimated 50 million people and 61,800 MW of electric load.
Did ENTSO-E attribute the 2025 blackout to a single cause?
No. It cites a combination of many interacting factors.
