Downtime Cost Statistics (2026): What an Hour of Outage Costs

A payment screen times out during a promotion. Your support queue starts filling, engineers pause other work, and the question stops being whether the incident is serious: what is each minute costing the business?

The short answer is that no universal average cost of downtime exists. The most useful 2026 downtime cost statistics range from a $15,000-per-minute average in Global 2000 research to a $2 million-per-hour median for high-impact outages—but those figures describe different populations, event definitions, and evidence types. Use the statistic whose scope resembles your decision; do not turn them into a single benchmark.

Key downtime cost statistics

• $600 billion annually: Cisco’s May 2026 release on Splunk research with Oxford Economics reports $600 billion in annual unplanned-downtime costs for Global 2000 companies, a modeled research finding that was 50% higher than two years earlier. The research also reports an average cost of $15,000 per minute for organizations in that Global 2000, cybersecurity-triggered downtime context. Oxford Economics says the 2026 research captured responses from 2,000 executives at Global 2000 companies. Cisco’s release and Oxford Economics’ research page describe the commissioned research.

• $95 million annually per organization: The same 2026 Global 2000 research reports $95 million in annual lost revenue per organization, nearly twice its 2024 level. This is an annual revenue-loss finding in commissioned research, not a price for a single outage.

• $2 million per hour: New Relic’s 2025 Observability Forecast reports a median cost of $2 million per hour for high-impact outages. ETR surveyed more than 1,700 IT and engineering teams and leaders across 23 countries and 11 industries in April and May 2025. “High-impact” is part of the metric; it is not an average for all outages. New Relic’s 2025 release also reports a $76 million annual median cost from high-impact IT outages among surveyed businesses.

• $14,056 per minute: A 2024 BigPanda page summarizing independently fielded Enterprise Management Associates (EMA) research reports an average cost of $14,056 per minute for an unplanned IT outage, nearly 10% higher than the 2022 figure. This is vendor-commissioned research, and EMA describes the result as defensible rather than definitive. See BigPanda’s research page and EMA’s description of the study.

• More than $100,000 per incident: In Uptime Institute’s 2025 analysis of its 2024 survey, 54% of respondents said their most recent qualifying significant, serious, or severe outage cost more than $100,000. This is a reported cost band for a respondent’s most recent qualifying outage, not an average cost. Uptime Institute’s annual outage analysis says one in five respondents put that most recent outage above $1 million.

• More than $300,000 per hour: ITIC’s 2024 survey of more than 1,000 firms worldwide, fielded from November 2023 through mid-March 2024, found that over 90% of mid-size and large enterprises reported an average cost of a single downtime hour above $300,000. ITIC says this self-reported exposure excludes litigation and civil or criminal penalties. ITIC’s Part 1 report carries the survey result.

• More than $100,000 per hour for 97% of large enterprises: ITIC’s 2024 Part 2 reports that 97% of large enterprises with more than 1,000 employees put their average single-hour downtime cost above $100,000. In the same survey family, 41% reported an hourly cost from $1 million to more than $5 million. These are self-reported exposure bands, not an industry-wide observed average. ITIC’s Part 2 report provides the large-enterprise breakdown.

• $400 billion annually, historically: Oxford Economics’ 2024 research, prepared with Splunk, estimated unexpected digital disruptions cost the Global 2000 $400 billion annually, or $200 million per company per year. That is a historical modeled annual estimate that includes market-capitalization and brand-reputation effects, so it is not comparable to an incident-level hourly median. Oxford Economics’ 2024 report page documents the earlier estimate.

Get the Mobile Testing Playbook Used by 800+ QA Teams

Discover 50+ battle-tested strategies to catch critical bugs before production and ship 5-star apps faster.

Email addressGet Instant Access

100% Free. No spam. Unsubscribe anytime.

Which downtime cost statistic should you use?

Choose by the question you need to answer, not by the largest number you can find.

The table is a map, not a conversion chart. A high-impact-outage median and an average unplanned IT-outage cost can both be useful without being interchangeable. Likewise, the Global 2000 annual estimate includes broader business consequences that an incident timer does not capture.

Why the average cost of downtime varies so much

The phrase “downtime cost” hides several different measures.

Event severity changes the result. New Relic’s number is explicitly for high-impact outages. A study focused on severe digital failures will naturally produce a different result from one that averages unplanned IT outages across its own defined population.

The unit changes the question. Per-minute and per-hour figures help you reason about an active incident. Annual-loss figures describe accumulated effects across a year. Revenue loss, ransomware payouts, regulatory fines, and stock-price movement are related business consequences, but they are not interchangeable line items.

The population changes the result. Global 2000 companies, large enterprises with more than 1,000 employees, and a broad collection of IT and engineering respondents do not have the same revenue base, systems, or exposure. Preserve the cohort when you quote a number.

The evidence type changes the confidence you can assign. ITIC reports self-reported exposure. Uptime reports what respondents said about their most recent qualifying outage. Oxford Economics and Splunk present a commissioned survey-plus-model view of annual impact. None is a ledger of every outage invoice.

That is also why you should not divide a published annual figure by an assumed number of downtime hours and label the result an average cost. It may be helpful as an internal scenario calculation, but it is your conversion—not a source-published statistic.

What the 2026 Global 2000 benchmark includes

The 2026 Oxford Economics/Splunk research is the clearest current annual-loss frame in this set. Beyond the $600 billion aggregate annual figure and $15,000-per-minute average, Cisco’s release reports $95 million in annual lost revenue per organization, an average 3.4% stock-price decline after a downtime event, $40 million average ransomware payouts, and $51 million in average regulatory fines per organization in the research context.

Those figures show why a downtime-cost discussion can extend beyond the engineering response. A service interruption may affect transactions immediately, then create recovery work, compliance exposure, and investor consequences that emerge later. They do not establish that every organization will incur each consequence, or that a particular outage carries all of them.

For a business-case conversation, the annual view is useful when your audience needs to understand strategic exposure. For incident response, it is too broad to substitute for your own service-level, revenue, and recovery data.

Hourly exposure and incident thresholds tell a different story

ITIC and Uptime Institute provide two ways to talk about the cost of a serious event without presenting a universal tariff.

ITIC’s survey asks organizations about their average cost for a single hour of downtime. Its result is particularly useful when you are comparing self-reported enterprise exposure by company size. The exclusions matter: ITIC says its more-than-$300,000-per-hour result does not include litigation or civil and criminal penalties.

Uptime’s analysis instead gives you thresholds for the most recent qualifying outage that respondents experienced. If you need to communicate the chance that a serious incident crosses a six- or seven-figure boundary, the 54%-above-$100,000 and one-in-five-above-$1-million findings are more faithful than recasting them as an average.

Neither source tells you the cost of a routine five-minute degradation. Their definitions select for material outages or enterprise-level exposure, which is exactly why their scope should travel with the number.

What these statistics do not measure for mobile apps

The public record does not provide a general mobile-app dollar cost per minute. The studies above concern enterprise IT, high-impact digital outages, data-center and infrastructure reliability, cyber-triggered disruption, or modeled Global 2000 losses. None establishes a benchmark for a broken mobile checkout, a failed iOS release, or an escaped Android defect.

Quash has no published first-party telemetry, recurring-bug dataset, customer quotation, or completed experiment that measures mobile-app downtime cost or the dollar cost of escaped mobile defects. That absence is important: a mobile-specific number would require a defined population, a stated failure type, a measurement period, and a transparent cost model rather than an inference from enterprise research.

You can use enterprise downtime figures as context when setting the case for prevention and faster detection. You should not present them as if they were measured mobile-app outcomes. If your release decisions hinge on that distinction, start collecting your own data: affected transactions or sessions, duration, support and engineering recovery time, refunds or credits, and any contractual or regulatory impact.

How to cite downtime cost statistics responsibly

A usable citation preserves the figure’s full shape:

1. Name the metric. Write “median cost per hour for high-impact outages,” not merely “downtime costs $2 million an hour.”

2. Keep the population. “Global 2000 companies” and “large enterprises with more than 1,000 employees” materially limit what a reader can infer.

3. State the period. Use the research publication year and, where available, the survey fielding period.

4. Label the evidence type. A survey response, a modeled estimate, and a reported incident threshold answer different questions.

5. Keep related costs separate. Revenue loss, ransomware payments, regulatory fines, and market reaction may belong in a disruption-cost discussion, but they are not all direct operational downtime costs.

This level of detail makes a number more useful in a board deck or budget memo because the next reader can decide whether the comparison is valid. It also prevents a common error: using a statistic about a high-impact outage as the average cost for every brief interruption.

Methodology and limitations

This report compares the primary pages published by the research owners, commissioners, and reporting organizations. The evidence includes vendor-sponsored research, commissioned research, and membership-funded industry research. Sponsorship does not make a figure unusable, but it does mean you should read the measure and population before generalizing it.

The evidence is also heterogeneous by design. New Relic reports a 2025 survey result for high-impact outages. ITIC reports 2024 self-reported hourly exposure. Uptime Institute reports 2025 analysis of a 2024 survey about respondents’ most recent qualifying outages. Oxford Economics and Splunk provide 2026 modeled annual effects for the Global 2000, while the $400 billion figure is the explicitly historical 2024 comparison.

Uptime Institute notes that outage-frequency, severity, and cost methodologies remain uncertain because transparency and reliable reporting are limited. That is a reason to retain definitions, not a reason to discard every figure. The responsible conclusion is narrower: there is no source-supported, universal average cost of downtime.

Conclusion

The best downtime cost statistics are decision tools, not interchangeable slogans. Use the 2026 Global 2000 research when you need to discuss annual enterprise exposure, a high-impact-outage median when you are assessing severe digital-service risk, and Uptime or ITIC findings when their incident threshold or enterprise cohort matches your audience.

For your own organization, the next step is to measure the costs the public studies cannot see: which services failed, how long customers were affected, what recovery consumed, and which revenue or obligations were actually at risk. That gives you a downtime number that belongs to your business instead of a borrowed one.