Microsoft Build 2025 was a whirlwind of excitement and tension. Protests broke out over Microsoft’s cloud contracts with Israel, sparking big questions about tech and ethics. In the middle of the chaos, a secret chat leaked, revealing Walmart’s plans to use Microsoft’s new AI tools. All eyes were on Microsoft’s security and cloud strategies, as tech, politics, and accidental drama collided under one roof.
What were the main controversies and highlights at Microsoft Build 2025?
Microsoft Build 2025 was marked by high-profile protests against Azure’s Israeli defense contracts, a major accidental leak revealing Walmart’s adoption of Entra Web and AI Gateway, and heightened scrutiny of AI security. These events spotlighted ethical debates and shifting priorities in Microsoft’s cloud and AI strategies.
Scene: Coffee, Code, and Chaos at Build
I’ve seen my share of developer conferences—neon lanyards, the scent of burnt espresso mingling with fresh PowerPoint slides—but Microsoft Build 2025 wasn’t just business as usual. You could almost taste the tension in the air, sharp as ozone before a thunderstorm. The event was meant to be a technophile’s utopia, but it turned into a palimpsest of innovation, protest, and, well, a dash of accidental drama.
So, what happened? On a Wednesday afternoon, as Neta Haiby and Sarah Bird—Microsoft’s head of security for AI and head of responsible AI, respectively—were evangelizing their latest AI security protocols, the session was stormed by Hossam Nasr and Vaniya Agrawal. Both ex-Microsoft, both with activist credentials longer than a Kubernetes cluster log. They weren’t there for swag. Instead, they demanded answers about Microsoft’s cloud contracts with the Israeli Ministry of Defense, their voices ricocheting across the conference hall like marbles in a tin can.
Security hustled in, the livestream was muted, and for a few minutes, the audience was left blinking amid a sensory vacuum: excited murmurs, then the sterile hush of silence forced by corporate damage control. It reminded me of last year’s Build, when I missed a critical demo because I got lost in a side hallway—only this time, the detour was deliberate, public, and far more consequential.
The Accidental Leak Heard Around the Enterprise
If you thought things would quiet down after the protest, you’d be mistaken. Just as the session rebooted—like an app after a particularly nasty crash—someone backstage slipped up. A confidential Microsoft Teams chat flashed onto the main screen for everyone in the room (and, for a few seconds, the livestream) to see. There it was: a candid exchange about Walmart’s imminent rollout of Microsoft’s Entra Web and AI Gateway.
I have to admit, I felt a pang of schadenfreude—who among us hasn’t accidentally shared the wrong screen? But this was no ordinary faux pas. “Walmart is ready to rock and roll with Entra Web and AI Gateway,” read one message from a Microsoft cloud solutions architect. Meanwhile, a Walmart AI engineer chimed in with, “Microsoft is WAY ahead of Google with AI security. We are excited to go down this path with you.” Oops.
The specifics mattered: Entra, Microsoft’s identity management suite, and AI Gateway, a hyperspectral tool for managing model deployment on Azure Databricks, are the crown jewels in Microsoft’s enterprise security offering. The leak was practically a masterclass in how not to handle sensitive client communications. I wondered, would there be fallout? Or would Walmart just shrug and keep moving, as if nothing had happened?
Protests, Politics, and the Perils of Partnership
The Build 2025 fracas wasn’t an isolated storm cloud. It was more like the leading edge of a squall line sweeping through tech’s biggest conferences. Earlier in the week, CEO Satya Nadella’s keynote was hijacked by protestors, as was a session led by Jay Parikh, Microsoft’s CoreAI EVP. Agrawal, one of the protestors from the AI security talk, already had a reputation for such disruptions—she’d previously interrupted the company’s 50th anniversary event with Bill Gates, Steve Ballmer, and Nadella all present. That’s chutzpah.
These activists, many of them current or ex-Microsoft employees, argue that the company’s cloud infrastructure—Azure, to be precise—has become an unwitting accomplice in geopolitical disputes, notably the ongoing conflict in Gaza. The “No Azure for Apartheid” movement has orchestrated several high-visibility protests, calling for Microsoft to untangle itself from contracts with the Israeli military.
Microsoft, predictably, pushed back. In a statement quoted by the Seattle Times, the company insisted its IMOD relationship was “a standard commercial contract.” They cited internal audits and a third-party review, claiming no evidence surfaced that Azure or its AI stack had been used to cause civilian harm or otherwise breach Microsoft’s AI Code of Conduct. I had to stop and ask myself: is that enough? Or is it just a smokescreen, a way to sidestep the murky ethics of dual-use technology?
Under the Hood: Why Entra and AI Gateway Matter
Let’s talk tech for a moment—because for all the Sturm und Drang, this is still a builder’s conference. Microsoft Entra isn’t your run-of-the-mill login system. It’s a sophisticated identity and access management suite that wraps user verification, policy enforcement, and privileged access in a chemical bond of AI-driven anomaly detection. In 2024 and 2025, Entra rolled out features like continuous security posture assessments and cloud-native policy granularity, which, frankly, gave Okta and Auth0 reason to sweat. You can get lost in the details on Microsoft’s blog—if you’re into that sort of thing.
And then there’s AI Gateway, a suite bolted onto Azure Databricks. It’s not just about pushing models to production; it’s about doing so with the kind of guardrails you’d expect in a nuclear power plant. Usage tracking