AWS Didn't Win Because It Was First. It Won Because of What It Built While Alone.

On March 14, 2006, Amazon flipped on a service it described, with no fanfare, as 'storage for the Internet.' It was called S3, and a developer could now park a file in Amazon's data center, address it with a few lines of code, and pay by the byte. The compute service everyone remembers — EC2 — didn't arrive until five months later, when the beta went live on August 24, 2006.¹⁹ Storage was the starting gun, not the headline. And for the next four to five years, AWS ran that race entirely alone: Microsoft did not launch Azure commercially until February 2010 — having previewed it only in late 2008 — and Google's cloud platform did not reach general availability until 2011.⁷¹¹

The official story is that AWS won the cloud because it got there first. That is the tidy version, and it is wrong in the way that matters. Both giants eventually showed up — and both have taken share since. If being early were self-sealing, that couldn't happen.

Here is the thesis a sharp friend could repeat: AWS's moat isn't the head start itself — it's the three structural things AWS poured into the ground while nobody was watching, each of which compounds. First arrival was the ignition. The engine is what got built during the quiet.

The integration debt nobody notices until they try to leave

When a company picks AWS today, it almost never picks one service. It picks the storage, then the database that sits next to it, then the queue between them, then the identity system that governs all three, then the monitoring that watches the lot. AWS now runs hundreds of integrated services, and the longer a customer stays, the more of them get wired together. None of that wiring is portable. Moving to a rival doesn't mean re-pointing one connection — it means unpicking years of accumulated integration, line by line. That is the difference between a switching cost and a switching catastrophe. The lock-in was never a contract clause. It was the customer's own architecture, hardening around AWS like sediment.

When your API becomes the industry's grammar

Here is the part rivals underestimated. The S3 interface — the specific way you ask the cloud to store and fetch a file — didn't just become popular. It became the reference standard. Code written against S3 in 2006 still runs today, and across the storage industry, competing vendors now build tools that speak the S3 dialect rather than invent their own.⁸ Think about what that means: even a company trying to escape AWS often does it on AWS's terms, because its engineers learned to think in S3's grammar and its tooling assumes it. An API standard is a quieter, deeper moat than any single product, because it locks in muscle memory. You don't compete with a language by building a better one — you compete by convincing the whole world to forget the one it already speaks.

The barrier that gets taller the more it's attacked

The third reinforcement is the most brutal and the least visible: money. AWS was a $4.6 billion business a decade ago; in fiscal 2024 it cleared $108 billion in revenue, up 19% year over year.³ And it isn't growth for growth's sake — AWS threw off roughly $39.8 billion in segment operating income in 2024 — against Amazon's total operating income of $68.6 billion, making AWS the majority contributor despite representing around 17% of its revenue.³¹⁰ That profit isn't a victory lap. It's ammunition. It funds the next wave of data centers, custom silicon, and global regions — the physical mass that a challenger has to replicate before it can even compete on equal footing. The barrier doesn't just exist; it grows, because the leader's profits buy more barrier faster than a rival's losses can buy a way around it.

If the moat is so deep, why is the share slipping?

The honest objection is that this all sounds too tidy — and the numbers half-agree. AWS's lead is being squeezed. On Gartner's IaaS-specific measure, AWS slid from 39% share in 2023 to 37.7% in 2024, while Microsoft edged up to 23.9% and Google to 9%.⁶ On Synergy Research's broader cloud-infrastructure measure, AWS sits closer to 31%, ahead of Azure's 20% and Google's 11%.⁵ However you cut it, the rivals who arrived years late are gaining ground, not losing it. So is the moat real?

It is — but a moat is a defense, not an immunity. The slow erosion is exactly what the strategic literature predicts: first-mover advantage converts to durable moat only when it's backed by switching costs, proprietary control, and learning-curve lock-in, and even then it holds the center while ceding the edges. AWS still leads by a margin no one else is close to. What the share drift proves is the opposite of the popular myth: the head start alone never protected anything. The integration debt, the API standard, and the capital wall are doing the protecting now — and they're holding a commanding lead against two of the best-funded companies on earth. That is what a working moat looks like under pressure, not the absence of one.

AWS is taught as the patron saint of first-mover advantage, and it's the wrong lesson. Plenty of companies got somewhere first and then watched the latecomers eat them. What separated AWS was not the calendar — it was the discipline to spend a quiet, uncontested half-decade laying down three things that keep paying out long after the head start was gone. First-mover advantage didn't build the moat. It just bought the time to dig it.