Ex-Microsoft engineer blames Azure problems on talent exodus • The Register

In a series of detailed essays, former Microsoft engineer Axel Rietschin has laid out a damning history of Azure's foundational problems, arguing that the cloud service's persistent issues stem from a rushed launch in 2008, a subsequent talent exodus, and years of under-investment in people rather than infrastructure.

Rietschin, who worked on Azure Core Compute for a year and as a Windows Base Kernel engineer for eight years before that, describes a platform that was pushed to market prematurely to compete with Amazon Web Services. "Azure never operated as smoothly or independently as promised," he wrote. "What Microsoft presented to the world, and to its most demanding customers, was a sophisticated system perpetually on life support."

The consequences of these rushed decisions have been building for years, according to Rietschin. Small but ongoing disruptions accumulated over time, creating what he calls "foundational fragility" in the platform. This fragility became visible in several high-profile incidents, including federal cybersecurity evaluators reportedly dismissing Microsoft 365 Government Community Cloud High (GCC High) as inadequate in 2024, and OpenAI's $11.9 billion compute deal with CoreWeave in March 2025, which Rietschin interprets as a vote of no confidence in Azure's ability to meet demanding requirements at scale.

Microsoft's workforce reductions during May-July 2025, which saw around 15,000 people laid off, further exacerbated the problem. Rietschin argues that these cuts came at a particularly inopportune time, as the company was already struggling with knowledge dilution caused by high attrition rates.

The current AI frenzy has only made matters worse, according to industry observers. Martin Alderson, co-founder of catchmetrics.io, warns about the "coming compute crunch" as AI coding agents dramatically increase the volume of code being produced and deployed. "With coding agents being able to output tens of thousands of lines of code, we're also seeing a massive spike in demand for compute on CI/CD workflows to test and deploy this code," Alderson explained.

This surge in AI-generated code is overwhelming infrastructure across the industry. GitHub, which is transitioning more of its services to Azure, has reportedly seen its uptime dip below 90 percent. While GitHub attributes some of its infrastructure challenges to its migration to Azure, Rietschin notes that it's unclear whether Azure itself is contributing to the instability, as the move may not yet be complete.

Rietschin's central argument is that the tech industry's under-investment in people – its willingness to discard experienced engineers – is being made worse by over-investment in AI. As more code is created, committed, and run on cloud services, the need for human oversight and infrastructure maintenance has never been greater.

Despite the current enthusiasm for AI replacing human developers, Rietschin remains skeptical. "LLMs are very good at reproducing patterns, so they help mostly when recreating variations of software that has been seen many times in the training set," he said. "They also help find bugs, not by 'understanding' but by observing deviations from their probabilistic expectations, again based on learned patterns."

He dismisses the notion that AI will replace software engineers as "sensationalism," arguing instead that Microsoft and other tech companies need to focus on bringing back senior technical leaders and investing in people through mentoring and coaching. "I think their most significant challenge was knowledge dilution caused by high attrition," Rietschin concluded, suggesting that the solution to Azure's problems lies not in more automation, but in better support for the human engineers who build and maintain these critical systems.