Case study · August 2016

A power switch failed. $150 million and three days followed.

A single piece of data center power equipment failed to switch over correctly, and Delta's backup systems didn't fully compensate. The result: roughly 2,000 cancelled flights and a multi-day recovery, despite the actual power failure lasting only hours.

Delta data center outage cost Delta 2016 flight cancellations airline data center failure backup power failure cost example

Read the quick facts Model an airline outage

Scroll for the timeline

2,300 Flights cancelled over three days

$150M Reduction to pre-tax income

Quick facts

What happened, in one table.

Sources are linked inline, drawn primarily from Delta's own SEC filings and investor disclosures.

Date August 8, 2016, with recovery extending through August 10.

What broke A critical piece of power control equipment at Delta's Atlanta data center failed, and some systems did not switch over to backup power as designed — a partial failure of the redundancy that data centers are built specifically to prevent.

Scale The outage forced Delta to cancel roughly 2,300 flights over three days, stranding passengers worldwide even though the underlying power issue was resolved within hours.

Recovery gap The core technical failure — a power switch — took hours to fix. Restoring the full airline operation, including aircraft and crew positioning, took days, illustrating how physical operations recovery can outlast the technical root cause by an order of magnitude.

Reported cost Delta disclosed in SEC filings that the outage and its recovery reduced pre-tax income by approximately $150 million, split between the immediate August revenue hit and additional recovery costs into the following quarter.

Why it cost so much

Redundancy that doesn't get tested isn't redundancy.

This incident is the clearest example on this list of the gap between "the outage is fixed" and "the business is recovered."

Backup power is only as good as its last test

Failover systems are designed for exactly this scenario, and still didn't fully activate — a reminder that redundancy needs to be tested under realistic failure conditions, not just verified as installed.

Airline operations recover slower than airline systems

Once systems were back online, aircraft and crews were still out of position across the network — the same cascading-schedule dynamic that makes airline outages consistently more expensive per hour than their IT root cause alone would suggest.

A short outage can still be a nine-figure one

The power failure itself was resolved in hours, yet the total cost reached $150 million — cost scales with operational cascade and recovery complexity, not just with how long the original fault lasted.

FAQ

Delta 2016, answered.

Questions that come up when citing this incident in a data center resilience or disaster recovery case.

Was this a cyberattack or a hardware failure? A hardware and power-system failure — Delta attributed it to an equipment malfunction in its data center's power infrastructure, not an external attack.

Why did it take three days to recover from an hours-long power issue? Cancelled and delayed flights left aircraft and crews out of position across Delta's network; rebuilding a legal, functioning schedule from that state takes substantially longer than restoring the underlying system.

Did Delta change its infrastructure afterward? Delta publicly committed to reviewing and strengthening its data center resilience and backup power testing following the incident.

How would this map to the calculator? The airline calculator models this cascading-delay dynamic directly — a short MTTR input still produces a large annual cost once frequency and cascade are factored in.

Your turn

What would a data center failure cost your operation?

Model your own systems, revenue, and recovery time using the same formula.

Open the airline calculator All case studies