Cloudy with a Chance of Disruption

Cyclic Disruption: A Series on Technology and Human Nature

The Situation

In the mid-2000s, I was a consultant working with multiple clients in financial services. Cloud was already in the air (pun intended), but only as an idea. For years people had tossed around terms like “utility computing” or “grid computing,” usually in conference talks or white papers that made it sound more fanciful than practical.

Then AWS arrived with S3 in 2006, and suddenly it wasn’t theoretical anymore. Here was a simple but profound offering — companies could store data in the cloud. The same kind of infrastructure that Amazon had built for itself was now open to anyone with a credit card. Soon after, Google followed, making its own infrastructure available to the world.

It felt like a turning point of sorts, but it also felt fragile and somewhat unsafe.

I remember one of the early S3 outages in July 2008, a single corrupted bit in a handful of messages caused corrupted system state information to spread throughout AWS’s distributed system through what they called a “gossip protocol” — the very mechanism designed to detect failures became the vector that propagated them, and for hours, data simply was completely inaccessible.

I watched the panic ripple through my friend’s startup; the team had bet on S3 for their core product. Apparently, their CEO was furiously calculating burn rate on a whiteboard while engineers hit refresh on the AWS status page every thirty seconds. The CTO, who’d championed the migration, sat silent in the corner — he knew he’d have to face the board, and they would want to know why the company’s entire operation could be frozen by someone else’s infrastructure.

That single outage carried more weight than months of successful uptime. And it reinforced the skepticism that was already thick in the air, especially in banking.

The Pattern Emerges

In the late 2010’s, many banks still resisted the idea of Cloud. I remember sitting in a risk committee meeting at a major financial institution where the head of infrastructure pointed out the window toward a building on the campus. “We spent $300 million building that data center,” he said. “You want me to tell the board we’re walking away?”

The conversations were remarkably consistent:

• “We’ll never put sensitive data outside our own walls.”
• “No regulator will approve this.”
• “Our infrastructure is too critical to entrust to someone else.”

Part of the resistance was economic, as banks had poured billions into their own data centers, and they weren’t eager to walk away.

Part of it was cultural — trust and security were non-negotiable, and no one wanted to be the first to fail.

And part of it was fresh in memory; the February 2017 S3 outage, caused by an engineer who mistyped a command during routine debugging and accidentally took down services for Quora, Slack, Docker, and hundreds of others for four hours. The incident cost S&P 500 companies an estimated $150 million.

Every outage became political ammunition. The people who’d argued against Cloud now had proof it was reckless, and the people who’d pushed for it were defending a decision that, in retrospect, looked naive.

Meanwhile, other industries were moving quickly. Startups built entirely on cloud. Netflix had completed its migration to AWS in early 2016 — seven years of methodically moving every aspect of its streaming infrastructure. Retailers and media companies found agility they’d never had before.

The contrast was pretty striking — while banks debated the risk, others quietly built their future on flexible, dynamic infrastructure.

Later came the quiet converts. Even while executives publicly resisted, development teams were running experiments. A test environment on EC2. A marketing site on cloud hosting. Small bets that, if they worked, could be scaled — and if they failed, could be quietly shut down before anyone noticed.

The Surprising Shift

Then in late 2016, Capital One did something completely unexpected; they announced they were moving everything to AWS. Not a pilot, not a hybrid strategy, but all infrastructure.

I remember reading the announcement and thinking This won’t work. The regulators will never allow it.

But they did. Over eight years, Capital One closed all eight data centers. They didn’t just move applications — they rebuilt 80% of them from scratch to be cloud-native. They went from quarterly releases to deploying code multiple times per day and hired an army of engineers who’d never worked in a traditional bank environment.

And the regulators? They didn’t block it. They worked with Capital One to figure out how to govern it, creating an early framework for other banks to begin considering a similar transition.

What struck me most was how little fanfare there was when other banks started following. No dramatic announcements, just quiet shifts. An SVP of infrastructure I’d known for years who’d spent a decade defending on-premises suddenly mentioned, casually, that they were “looking at Cloud for some workloads.” Six months later, they had a migration roadmap. A year after that, they were further along than anyone admitted publicly.

Job postings changed, and suddenly every senior role required AWS certification. The debates in risk committees shifted from “whether” to “how fast.” The same executives who’d called it reckless were now worried about being left behind.

There was no single day when the banking world declared itself “in the Cloud.” It crept in, workload by workload, until one morning everyone realized the decision had already been made.

A Familiar Reflex

I’ve watched this pattern before.

In the 1980s, I saw finance teams resist spreadsheets with the same intensity banks later resisted cloud. VisiCalc, then Lotus 1-2-3 — they were toys. “Show me the formulas,” controllers would say. “Where’s the audit trail?” They’d point to their ledgers, their calculators, their proven methods.

Then someone used a spreadsheet to finish a quarterly close in three days instead of two weeks. Someone else built a budget model that could test twenty scenarios in an afternoon. Suddenly the person still using ledgers looked slow, not careful.

Within five years, those same controllers were building their entire planning processes in Lotus, and the tool they’d called unreliable became the thing they couldn’t imagine working without.

That’s what I keep coming back to… the refusal always feels justified in the moment. The risks are real. The track record is thin. The people advocating for change often don’t understand the complexities of the existing system.

And yet, once the economics line up and someone demonstrates the advantage, the resistance doesn’t fade gradually. It collapses. What was unthinkable becomes standard so fast that we forget we ever argued about it.

Why This Matters Now

Just two days ago, on Monday, October 20, 2025, AWS suffered another major outage. For hours, services from Starbucks to Snapchat to Fortnite to Coinbase to The New York Times went dark; myriad enterprises were impacted behind the scenes. Tens of thousands of users reported problems, and some businesses lost hundreds of thousands of dollars per hour.

And yet, the response was remarkably muted.

There was frustration, yes. Social media filled with complaints. But there was no fundamental questioning of whether Cloud itself was viable. No executives declaring they’d move back to on-premises infrastructure. Instead, the conversation was about redundancy strategies, multi-region deployments, and disaster recovery planning.

The outage that would have validated every skeptic’s fears in 2008 barely registered as a strategic question in 2025. We’ve already crossed the threshold where cloud is simply how things work.

For me, the most interesting part of the cloud story isn’t just the resistance. It’s how the shift happened almost invisibly. Resistance always feels loud in the moment. But when the adoption comes, it comes so quietly and thoroughly that we barely remember the fight.

I hear echoes of that today. Leaders dismissing intelligent systems as overhyped or untrustworthy. Teams saying “we’ll never use this for anything critical.” Executives pointing to the mistakes, the failures, the risks.

I’ve heard it before.  You have, too. We’ll see how it plays out.

Your Next Encounter

The next time you hear someone say “this will never happen here,” pause and remember how many times we’ve said the same thing.

• Remember the banks that spent billions on data centers, certain that cloud was too risky for financial services.
• Remember the CFOs who trusted ledgers more than spreadsheets.
• Remember the executives who declared that their sensitive data would never leave their walls — right before they hired a team to plan the migration.

By the time it feels inevitable, the advantage will already belong to those who moved while others resisted.

Previous Post: Declaring Agile by Fiat

Next in this series: The Illusion of Control — Why automation creates comfort before it creates clarity.

About this series: Cyclic Disruption explores patterns in how humans adapt to transformative technology, drawn from four decades of experience in development, consulting, and leadership. Each essay connects a past moment to today’s dilemmas — showing what we can learn when we stop treating our own creations as strangers.