Six weeks, six failures. One shared risk.

The Backhoe and the Broken Promise of Redundancy

On May 5, a contractor cut a fiber line in western Pennsylvania. Within the hour, Verizon customers across the eastern third of the country were down - Mahoning Valley, New York, Atlanta, New Orleans, parts of Florida. Field-service apps stopped syncing. Calls dropped outright. Full restoration took nearly seven hours.

Carriers use the word redundancy in every proposal and every SLA. What they often don't say is that redundant routes frequently share the same physical conduit and the same long-haul corridors. A single shovel can take down what was sold as three separate paths, because two of them ran through the same trench. The logical topology in the portal looks fine. The physical reality is that almost nobody on the carrier side verifies true diversity end to end.

The fix starts with asking your carrier, in writing, for the entrance facility and long-haul corridor each circuit actually travels. If they can't answer, that's your answer.

The Cloud Is a Building, and Buildings Break

Three days later, on May 8, the air conditioning failed in an AWS data center in Northern Virginia. Servers shut themselves down to keep from overheating. Coinbase went offline for more than five hours. FanDuel went dark mid-trading. Thousands of smaller Midwest businesses that didn't know they depended on US-EAST-1 found their scheduling platforms, payroll, time clocks, CRMs, and cloud point-of-sale stumbling at once.

Ohio should pay attention here. Central Ohio has become a major data center market, and the same concentration that took down US-EAST-1 is being built into the grid around Columbus right now. For a multi-location business, the real question is no longer whether the cloud is reliable. It's whether you know which buildings your business actually runs out of, and what happens at your sites when one of those buildings fails.

The Copper Sunset Is Happening to You

While outages made headlines, a slower deadline kept advancing. AT&T is shutting down copper wire centers starting in June 2026, about 500 in the first wave, with Ohio on the list. By November 15, a lot of POTS lines stop working.

If your building has a fire alarm panel, an elevator emergency phone, or a security panel that calls a monitoring station, there's a decent chance it still rides analog copper. When that line goes dark, the panel can't reach the monitoring center, and moving it to VoIP is not automatic - most fire panels won't pass UL864 over a standard SIP trunk, and the local authority having jurisdiction has to sign off too.

Businesses that called their carrier in March were quoted June install dates. Businesses that waited until April are now looking at September, past the cutoff. Replacement hardware is part of that queue, so the order date matters as much as the install date.

Rural Ohio Is Paying for the AI Buildout

Meta signed a large fiber supply deal with Corning, and the fiber supply chain shifted overnight. Midwest ISPs that won BEAD awards told Light Reading that fiber orders were canceled, and industry sources there tied the cancellations to the Meta deal. One of those ISPs was told the loose tube fiber it would normally buy from Corning, the kind rural fiber-to-the-home needs, now has a 52-week lead time. Superior Essex told Fierce Network that ribbon fiber lead times are over 60 weeks. Smaller providers report steep cost increases.

If you operate in rural Ohio - manufacturers in Wayne County, healthcare networks in the southeast, distribution out of Findlay or Mansfield - that supply squeeze hits directly. Buildouts slip a year. ISP partners walk away when the economics no longer work. The honest answer right now is fixed wireless, hybrid topologies, microtrenching, and leased dark fiber from cooperatives that planned ahead - not a build date nobody can hit.

UCaaS Cutovers and the Vendor Chain You Didn't Sign Up For

The second category of concentration risk looks like a project plan, not a fiber cut. A twelve-week cloud phone cutover across a few sites turns into month seven. Porting dates slip. E911 records carry addresses that haven't been accurate in years, so a 911 call from a back warehouse gets routed to the wrong dispatcher. Nobody tested the handsets on the real network under real traffic until cutover day. The platforms work fine when someone who knows what they're doing runs the project - the failure is a single national provider running a shared queue of project managers across dozens of accounts, none of whom has seen your network.

The same pattern shows up in hosting. In January, a Columbus hosting provider went dark for 76 minutes and took several unrelated downstream companies down with it. When you sign with a large carrier or national MSP, you're really signing with a chain: an ISP, a hosting platform, hardware suppliers, cloud platforms - any one of which can take you down without warning. The question worth asking isn't how big your provider is. It's whether there's one number to call when something breaks, and whether the person who answers has the authority to fix it.