Azure Was Built Under Ballmer. Nadella's Genius Was Wiring It Into Everything.

On February 1, 2010, Microsoft quietly switched on a thing called Windows Azure. It was not an infrastructure cloud you could rent servers from; it was a Platform-as-a-Service offering - a place to run code, not racks.² The CEO was Steve Ballmer. The strategist who had unveiled it sixteen months earlier at a developer conference was Ray Ozzie, who described it as 'a service, hosted and maintained by Microsoft on an array of distributed data centers.'² The internal code name had been 'Project Red Dog.'¹ Satya Nadella would not be CEO for another four years. Hold that date in your head, because almost every story you've heard about Azure gets it wrong.

The official story is that Nadella invented the cloud at Microsoft - walked in on day one, said 'mobile-first, cloud-first,' and built the growth engine that revived a dying company. Almost none of that is true in the order people tell it. The platform predates him. The slogan came on day 52, not day one.⁶ What Nadella actually did was harder and less photogenic than invention: he took an existing PaaS experiment and turned it into the most valuable reporting segment Microsoft has.

Azure didn't need to be invented. It needed to be repositioned.

Here is the move people miss. Ballmer-era Azure was a developer's sandbox for running Microsoft-flavored code - a logical extension of the Windows franchise into the cloud. That framing was also its ceiling. A PaaS that assumed you were already a Microsoft shop could only ever sell to people who already bought Microsoft. The whole rest of the market - the Linux servers, the open-source databases, the startups that would never touch a Windows license - was structurally locked out by the product's own design. Microsoft had spent a decade treating open source as the enemy. The cloud business inherited that posture, and it was strangling demand at the door.

The repositioning was to make Azure indifferent to what you were carrying. Embrace Linux. Run the open-source stacks. Stop selling 'the Windows cloud' and start selling 'the cloud,' then let the rest of Microsoft's catalog ride in behind it. This is the part the slogan obscures. 'Mobile-first, cloud-first' sounds like a bet on a new product. It was really a decision to stop defending the old one - to let the cloud be neutral ground so that the enormous installed base of Office, Windows Server, SQL Server, and eventually GitHub had somewhere to migrate without being asked to convert.

The number Microsoft will sell you, and the one it won't

Try to find Azure's revenue in an SEC filing and you can't. Microsoft does not report it as a standalone line.⁸ What it reports is the Intelligent Cloud segment - Azure bundled together with SQL Server, Windows Server, GitHub, and a stack of other services - and that segment is the tell. In FY2020 it was $48.4 billion.⁴ By FY2024 it was $105.4 billion - roughly a 118% jump in four years.³ Against total company revenue of $245.1 billion and operating income above $109 billion that year, the cloud segment was the primary driver of the whole machine.⁵ Executives have said Azure and cloud services passed $75 billion in FY2025 on earnings calls - but there is no audited standalone Azure P&L, only growth rates disclosed inside the segment.⁸

The bundling is not an accounting accident; it's the strategy in financial form. By reporting Azure inside a segment with the Server products it cross-sells, Microsoft makes the migration path part of the story. The reason a number like 'Azure's exact revenue' doesn't exist is that Microsoft doesn't run Azure as a separate thing. It runs Azure as the gravity well into which the rest of its catalog falls.

Isn't 'second to AWS' just a polite way of saying it lost?

The fair objection is that Azure didn't win - AWS got to the cloud first, defined Infrastructure-as-a-Service, and still leads. Even the share figures bear that out, sort of: Canalys put Azure at 20% on a spend basis in Q2 2024, HG Insights at 24% on an install-base basis in Q1 2024, both behind AWS at roughly 31-33%.⁷ So the honest read is that Azure is a strong number two, not a conqueror. But notice the figures themselves prove the deeper point: they disagree because analysts measure different things, and there is no single authoritative percentage. The interesting fact isn't the share - it's that Microsoft turned a late-arriving, Windows-flavored PaaS into a contender at all, by refusing to let the cloud stay a Windows product. Being a durable number two on a market this size, attached to the world's most entrenched enterprise software catalog, is not the consolation prize. It's the prize.

So the tidy origin myth has the verbs backwards. Azure was not invented in 2014; it was launched in 2010, under a different CEO, with a code name about a dog.¹ What changed was not the platform's existence but its posture - from a cloud that asked you to already be a Microsoft customer to a cloud that took your Linux box and quietly sold you the rest of Microsoft over the following decade. The slogan got the headlines. The repositioning got the $105.4 billion segment.³ The lesson sits uncomfortably with how we like to tell business stories: the growth engine was already in the garage. Someone just had to stop guarding the door it was parked behind.