What If the Internet Went Down for a Week?

A week-long global internet blackout is a different animal than a permanent one. The moment people know it will come back — on day 7, not never — every decision they make changes: businesses ration cash instead of collapsing, hospitals ration bandwidth instead of reverting entirely to paper, and governments manage a crisis instead of rebuilding civilization. This piece walks through exactly what a bounded, known-duration outage would do to payments, supply chains, hospitals, and emergency services, using real historical outages — Egypt's five-day shutdown in 2011, Mauritania's cable-cut blackouts, and the 2024 Red Sea cable cuts — as the closest evidence we have.

Thought experiment. This article explores a hypothetical scenario for educational and speculative purposes. The events described are not predictions and should not be interpreted as warnings, advocacy, or instructions.

Why "one week" is a meaningfully different scenario

Most internet-collapse thought experiments treat duration as a footnote. It isn't. A permanent loss of the internet forces total reinvention of how an economy runs. A one-week loss, by contrast, is something humans have actually lived through in bounded, regional form, and the data from those events tells a very different story: the damage is front-loaded, the adaptation is fast and improvised, and the deepest costs come from irreversible failures that happen to fall inside the seven-day window (a missed harvest shipment, a patient who needed same-day chemotherapy scheduling, a business that couldn't make payroll and folded) rather than from the outage itself grinding everything to zero.

That distinction matters for readers comparing this scenario to a permanent internet loss. If you've read our companion piece on the internet disappearing forever, you'll notice the first 72 hours look almost identical there and here. The two scenarios diverge sharply after day 3, once the "when does this end" question starts shaping behavior. A known one-week clock lets institutions triage: prioritize the systems that must survive the week, let non-critical services sit dark, and hold on. A permanent outage offers no such luxury — everything has to be rebuilt from scratch, which is why that scenario ends in a described retreat toward pre-digital infrastructure rather than a recovery curve.

Real-world precedent for the "known, finite duration" version of this scenario is thin but real. Governments have deliberately cut national internet access for days at a time, and undersea cable damage has knocked out regional connectivity for similar windows, giving us actual data instead of pure speculation.

When Egypt's government ordered internet and mobile-data providers offline for five days during the January 2011 uprising, the Organisation for Economic Co-operation and Development estimated the direct cost at roughly $90 million — about $18 million per day for one mid-sized national economy. A global, week-long blackout would multiply that scale by orders of magnitude across every connected economy on Earth simultaneously.

Real-world case studies: what finite outages actually looked like

Three historical events come closest to modeling a "the internet goes dark, then comes back" scenario, each at a different scale.

Egypt, January 27 to February 2, 2011. Facing nationwide protests, the Mubarak government ordered domestic ISPs and mobile carriers to cut service. Egypt effectively vanished from the global internet for five days. The OECD's post-event analysis put the direct economic cost at $90 million, driven mainly by lost e-commerce, tourism booking disruption, and stalled outsourcing and call-center work that Egypt had built as an export industry. Voice calls over the old fixed-line telephone network kept functioning throughout, which is a large part of why the country didn't see the kind of emergency-response failures a total communications blackout would cause.

Mauritania, March 2018 and again in 2024. A break in the ACE (Africa Coast to Europe) submarine cable near Nouakchott cut off Mauritania's internet access entirely for about 48 hours in 2018, and a separate cable fault caused another multi-day disruption in 2024. Mauritania depended on a single cable for essentially all of its international connectivity — a single point of failure that also knocked out Sierra Leone, Liberia, Guinea-Bissau, Guinea, and the Gambia, which shared the same cable. Repair required a specialized cable-laying ship to physically locate and splice the break at sea, which is why even a short-duration cause (a ship's anchor, a seabed shift) produces a multi-day-to-multi-week fix.

The Red Sea cable cuts, February–March 2024. Multiple major submarine cable systems — SEA-ME-WE-4, IMEWE, EIG, and FALCON GCX — were damaged in the same narrow chokepoint near the Bab-el-Mandeb Strait, likely from a ship's anchor dragging across the seabed after a Houthi attack in the area. Estimates of the traffic impact ranged from roughly 25% up to nearly 70% of Europe-to-Asia data flow, depending on which routes and providers were measured. Traffic that would normally cross the Red Sea had to reroute around Africa or through alternate cable systems, adding latency and congestion for weeks even though no single country went fully dark.

These three events share a pattern relevant to a global one-week scenario: the initial shock (hours 1 to 48) causes the sharpest behavioral disruption, connectivity workarounds emerge fast (rerouted traffic, cash economies, radio and landline fallback), and the deepest economic damage accumulates from businesses and processes that simply couldn't wait out the clock, not from the outage itself worsening over time.

Hour zero to day one: the shock phase

The first 24 hours of a global blackout produce panic and confusion more than structural collapse, because most physical infrastructure has enough local buffer to run on autopilot for a day.

What keeps working:

• GPS navigation, which runs on dedicated satellite constellations (GPS, Galileo, BeiDou, GLONASS) that are architecturally separate from the internet.
• Terrestrial broadcast TV and radio, along with satellite TV.
• Traditional landline (PSTN) telephone service, still functional in most countries even if thinning.
• Cellular voice calls and SMS, which ride on carrier-managed circuit and signaling networks distinct from mobile data.
• The power grid, at least regionally — plants can run on local manual control even if remote SCADA monitoring drops offline.

What breaks almost immediately:

• Card payments. Visa, Mastercard, and mobile wallets all require real-time online authorization, so point-of-sale terminals stop working within minutes. Stores go cash-only, cash-only stores run out of small bills fast, or businesses simply close.
• ATM withdrawals, since most machines need to check balances against a central server before dispensing cash.
• Ride-hailing, food delivery, and e-commerce, which are internet-native by design.
• Office collaboration tools — Slack, Microsoft Teams, Google Workspace, enterprise CRM systems — all go dark simultaneously.

Stock exchanges close within hours because modern trading infrastructure cannot function without networked order matching, clearing, and settlement. For scale, the 2015 NYSE technical outage — a single exchange, four hours, one cause — cost roughly $1 billion in disrupted trading volume; a global, week-long closure of every major exchange would be an unprecedented, if temporary, halt to price discovery across every asset class.

How would emergency services hold up during a one-week outage?

Emergency call routing increasingly depends on VoIP and internet-connected dispatch software, so a global outage would degrade 911/112-style systems in many regions within the first day. Analog radio dispatch and landline call centers, where they still exist, would carry most of the load. Response times would lengthen, particularly in jurisdictions that had already retired backup analog systems.

Days two and three: supply chains start to crack

Modern logistics runs on just-in-time delivery, meaning inventory buffers are measured in days, not weeks. Toyota and most large manufacturers typically hold one to three days of parts inventory. Trucking fleets dispatch through real-time routing software, and container ports schedule cranes through cloud-connected logistics platforms.

By day three of a global outage, the visible consequences stack up quickly:

System	Day 1 status	Day 3 status	Why it breaks on this timeline
Grocery restocking	Slight delay	Shelves visibly thinning	1–3 day just-in-time delivery cycles for perishables
Fuel stations	Some closures	Widespread closure	Pumps require online payment authorization and tank telemetry
Bank branches	Open, limited service	Withdrawal limits, some closures	Cannot verify account balances against central ledgers
Air travel	Delays	Near-halt	Scheduling, check-in, and routing run on cloud systems
Shipping/customs	Manual tracking begins	Containers untracked, customs backlog	Cloud-based crane scheduling and customs clearance offline

Table: how core logistics and financial systems degrade over the first three days of a global internet outage.

Fuel is a particularly sharp choke point. Even cash-paying customers hit closed stations, because most modern pumps cannot ring up a sale at all without their point-of-sale system reaching a payment processor, regardless of payment method. That single dependency — a $20 network authorization call — is enough to stop a fuel supply chain that otherwise has nothing to do with software.

Airlines illustrate how fragile scheduling software has already become even without a global outage. The 2022 Southwest Airlines crew-scheduling system failure — a single company's internal software problem, not an internet-wide event — stranded roughly two million passengers over about a week. A global internet outage would do that to every airline simultaneously, because none of them can schedule crews, check in passengers, or coordinate air traffic support systems without connectivity.

The Bank for International Settlements has modeled scenarios in which payment-network outages lasting five days or longer create conditions similar to a bank run, as depositors who cannot verify their balances or move money digitally rush physical branches for cash. In a week-long global scenario, that dynamic would likely play out simultaneously across dozens of countries.

Days four through seven: the systems built without a manual fallback

By the second half of the week, the outage stops being an inconvenience and starts exposing which critical systems were built with an offline fallback and which weren't.

Hospitals and healthcare

Electronic health record systems become inaccessible, forcing a hard reversion to paper charts in facilities that no longer keep paper as a routine backup. Electronic prescribing fails, so pharmacies cannot verify what a doctor actually ordered. Picture archiving and communication systems (PACS), which most modern imaging equipment relies on for storage and access, degrade because they're built around networked storage rather than local drives.

A useful real-world proxy: the 2024 ransomware attack on Synnovis, a pathology-services provider for the UK's National Health Service, took critical lab and blood-testing systems offline and delayed more than 10,000 procedures and appointments over roughly a week, according to NHS incident reporting. That was one supplier, in one country. A global internet outage generalizes that failure mode across every hospital system that depends on networked lab and imaging infrastructure simultaneously — which, in 2026, is nearly all of them in high-income countries.

Power and water utilities

Many electric utilities use cloud-connected SCADA (supervisory control and data acquisition) systems for monitoring and load balancing across regions. Individual plants can typically run on local, manual control for days without major issues, but coordination across plants and regions — balancing supply and demand across a national grid — degrades as smart-grid data stops flowing. That raises the risk of localized cascading blackouts, particularly during periods of unusually high or low demand. Water treatment plants generally have manual operational fallback, but automated chemical dosing logs and regulatory compliance reporting break down, which creates downstream paperwork and safety-verification problems even if water quality itself holds up.

Financial and economic disruption

Stock markets remain closed for the duration. Insurance claims cannot be filed or processed. Payroll systems, most of which now run through cloud-based providers rather than in-house software, fail — meaning millions of workers could miss a scheduled paycheck if the outage falls across a pay period. Allianz's cyber-risk research has estimated the cost of a major cloud or network outage at somewhere between $10 billion and $50 billion per day for large economies; extrapolated across a full week and across every connected economy simultaneously, the global total plausibly lands in the hundreds of billions to over a trillion dollars, a scale comparable to a significant financial crisis or major natural disaster.

Social and psychological effects

Roughly 70% of adults in developed economies report that digital channels are their primary way of communicating with friends and family, which means a week-long blackout produces genuine social isolation on top of the economic disruption. Misinformation tends to fill the vacuum left by official channels going silent — a pattern documented in multiple regional internet-shutdown studies, where rumor and word-of-mouth misinformation spread faster in the absence of verified digital news. Crime patterns shift too, since many modern surveillance systems, digital police records, and license-plate-reader networks depend on connectivity that simply isn't there.

What breaks permanently versus what just pauses

The single most important distinction in a bounded one-week outage, compared to a permanent one, is that most systems here are paused, not destroyed. Bank ledgers still exist; they're just unreachable for a week. Hospital records still exist on servers; doctors just can't see them. That distinction is why recovery, while messy, is measured in days and weeks rather than a full societal rebuild.

But some losses genuinely cannot be undone once the week has passed, regardless of how completely the network comes back:

• In-flight financial trades and transactions that were mid-process when the outage began. Reconciling exactly what was owed to whom, at what price, at the exact millisecond of failure, is a manual audit problem that can take financial institutions months to fully resolve.
• Missed medical procedures and appointments that depended on real-time scheduling, particularly time-sensitive treatments like chemotherapy cycles or organ transplant coordination, where a week's delay isn't just an inconvenience but a clinical setback.
• Perishable inventory losses across the food supply chain — produce, dairy, and other short-shelf-life goods that couldn't be shipped, sold, or refrigerated properly during the disruption.
• Businesses that fail during the outage window itself, particularly small businesses operating on thin cash margins that couldn't process any digital payments for seven straight days.

Recovery itself is technically nontrivial even once connectivity nominally returns. DNS caches, BGP (Border Gateway Protocol) routing tables, and TLS security certificates all need to repropagate across the network before services actually function normally again — restoring a connection isn't a single switch-flip, it's thousands of interlocking handshakes re-establishing simultaneously. Cloud providers need to verify data integrity before reopening services to avoid corrupting records that were mid-write when the outage hit. And reconnecting billions of devices at once creates what network engineers call a "thundering herd" — a surge of simultaneous reconnection requests large enough to re-crash the very systems trying to come back online.

Two real events illustrate this recovery lag even at much smaller scale. The October 2021 Facebook/Meta outage lasted about six hours from a BGP configuration error, but reconnecting roughly three billion users hammered the company's servers for additional hours after the underlying issue was fixed. The July 2024 CrowdStrike incident, which stemmed from a faulty software update rather than a network failure, still affected roughly 8.5 million Windows machines worldwide; aviation, hospitals, and banks needed days of manual recovery work even though the root cause was patched within hours.

Q: How long would it take for everything to fully recover after a one-week global outage ends? A: Based on recovery patterns from smaller real-world outages, expect core internet backbone connectivity within the first 24 hours, major cloud and financial settlement systems within one to three days, and full enterprise and government service restoration by roughly day seven post-recovery. Full data reconciliation — resolving corrupted transactions, missed reporting, and financial audits — can take weeks to months longer.

Which countries would suffer most, and why

Not every country would experience a week-long global outage the same way. The determining factor isn't wealth alone, it's how completely a country's payment and service infrastructure has already gone cashless and cloud-dependent.

Near-cashless societies like Sweden and South Korea, where a large share of daily transactions already happen digitally and physical cash use has declined sharply, would face the steepest immediate disruption to everyday commerce, precisely because their populations and retailers have less practiced fallback capacity. Countries that retain robust cash economies and manual banking processes, often in lower-income regions, would in some ways adapt faster to the payments side of the crisis, even as they face other serious challenges around supply chains and healthcare.

Estonia offers an interesting counterexample on the government-services side: its digital-first e-government system, one of the most advanced in the world, has been built with explicit offline contingency drills as part of national digital-resilience planning, since Estonia has direct historical experience with Russian-linked cyberattacks disrupting its networks in 2007. Very few other countries have built equivalent offline fallback drills into critical government infrastructure.

What people get wrong about this scenario

Myth: the whole internet could plausibly go down at once from a single cause. The internet's core design — multiple redundant paths between networks, thousands of independent internet service providers, and distributed routing — makes a true simultaneous global failure extremely improbable from any single technical cause. What the historical record actually shows is regional and national outages: government-ordered shutdowns like Egypt's, cable-damage events like Mauritania's and the Red Sea cuts, and company-specific software failures like CrowdStrike's. This article treats a global blackout as a thought experiment specifically because it lets us stress-test dependency, not because it's a likely event.

Myth: cell phones would go completely dark too. Voice calls and SMS text messages run over carrier-managed circuit-switched and signaling infrastructure that is largely separate from the public internet, so both would likely keep functioning even during a full internet outage. What fails is mobile data — meaning apps like WhatsApp, iMessage, and Signal, along with anything requiring a data connection, would stop working even though a basic phone call or text might still go through.

Myth: GPS and the internet are the same system. GPS satellites broadcast timing and positioning signals completely independently of the internet, so navigation would keep working throughout a full internet blackout. What breaks is anything layered on top of GPS that needs live data — traffic conditions, ride-hailing dispatch, real-time map updates — even though the raw positioning signal itself stays online.

Myth: a week-long outage and a permanent outage would look basically the same for the first few days. They actually would look similar at first, which is part of why the two scenarios get confused. The divergence shows up in decision-making, not in the visible chaos: institutions facing a known seven-day clock triage differently than institutions facing an open-ended collapse, prioritizing what has to survive the week rather than rebuilding from scratch.

Q: Is it worth keeping cash on hand because of scenarios like this? A: Financial-preparedness guidance commonly recommends keeping some cash and a week or two of essential supplies on hand as a general hedge against a range of disruptions, not internet outages specifically, since storms, earthquakes, and regional power failures create similar short-term access problems to banking and retail.

Q: What's the longest real internet outage anyone has actually experienced? A: Deliberate, government-ordered national shutdowns have run far longer than a week in some cases — internet access in parts of Myanmar and Iran has been cut for extended periods during periods of unrest. Among accidental, non-deliberate disruptions, the July 2024 CrowdStrike incident is among the most disruptive global-scale events on record, though it affected specific software rather than internet connectivity itself.

The takeaway

A one-week global internet outage would not end civilization the way a permanent loss might; it would instead act as a stress test that exposes exactly which systems were built with an offline fallback and which weren't. The pattern from real bounded outages — Egypt in 2011, Mauritania's cable cuts, the 2024 Red Sea disruptions — is that damage concentrates in the first 72 hours and in irreversible, time-sensitive failures (missed medical procedures, failed small businesses, spoiled perishable inventory) rather than in a steadily worsening collapse. The deeper lesson isn't that the internet is fragile in some abstract sense. It's that a surprising number of physically simple transactions, like buying gasoline with cash, now silently depend on a network authorization call that most people never think about until it fails.

---

Read next:

• What If the Internet Disappeared Tomorrow? (Permanent Loss Scenario)
• Why Undersea Cables Carry Most Internet Traffic
• How the Internet Works — From Click to Server
• How Satellite Internet Reduces Latency
• How Digital Signatures Prove A File Is Authentic

Sources

• OECD/Forbes analysis: economic cost of Egypt's 2011 internet shutdown
• Gulf News: Egypt internet shutdown cost $90 million, OECD says
• Fortune: Mauritania's entire internet cut off for two days after undersea cable break, 2018
• Semafor: damaged undersea cables disrupting Africa's internet, 2024
• CNN Business: Red Sea submarine cable damage disrupts internet traffic, March 2024
• Telecoms.com: Red Sea cable cuts' impact was severely underestimated
• Allianz Global Corporate & Specialty (2019). Cyber Risk Trends — Cloud Outage Scenarios.
• Bank for International Settlements (2023). Cyber resilience of the financial system.
• UK NHS Synnovis Pathology Incident Report, 2024.
• CISA Joint Advisory on the CrowdStrike July 2024 incident.
• Cloudflare Engineering Blog — analysis of the Facebook/Meta BGP outage, October 2021.