GitHub

On Tuesday, at Microsoft's annual Build conference, the company announced a series of new or expanded AI initiatives, including a super app, internal reasoning models, a cybersecurity tool, and OpenClaw-like AI agents. All of this news adds up to a clear message: Microsoft is positioned to be one of the biggest players in AI, and it's finally acting on it.

Microsoft and OpenAI have parted ways: they are now ready to fight

“We have to prove that we can do everything we need from the ground up,” said AI chief Mustafa Suleyman.

Microsoft and OpenAI have parted ways – now they're ready to fight

“We have to prove that we can do everything we need from the ground up,” said AI chief Mustafa Suleyman.

For years, Microsoft's AI business relied heavily on its early and exclusive partnership with OpenAI. But the marriage drama slowly turned into a situational situation, and the couple actually split in late April (although Microsoft is still OpenAI's primary cloud partner – for now). This year's Build had the vibe of a newly single divorcee posting a thirst trap on Instagram. "It's always fun to attend developer conferences during times of big change," Microsoft CEO Satya Nadella said on stage Tuesday, adding that events like this are about "capturing this new opportunity."

AI chief Mustafa Suleyman, in an interview with The Verge, put it even more bluntly.

“The goal is to prove that we can become one of the top four laboratories in the world,” Suleyman said. "There are three labs that matter, Google DeepMind, OpenAI and Anthropic. We're not one of them at the moment, and that's always been my intention. That's why I came here. I want to build the best border models in the world, fully multimodal, and to do that, we need to prove that we can do everything we need to from scratch, and we're not just going to take on others.

One of Microsoft's first steps during Build was to catch up on AI models. Suleyman unveiled MAI-Thinking-1, the company's first reasoning model, along with six other new models focused on image, voice, transcription and coding. Microsoft said the mid-size MAI-Thinking-1 model, which will likely be marketed primarily to enterprise customers, is "built from the ground up for serious math, coding, and real-world enterprise deployment." Microsoft is years behind OpenAI and Anthropic here; OpenAI began releasing reasoning models in fall 2024. But Suleyman highlighted its performance on criteria such as coding and its price, saying it was cheaper than its OpenAI equivalents on some tasks — a big problem in the era of the AI financial crisis, which has inspired many customer complaints.

While Microsoft has had years to learn from OpenAI, Suleyman was careful to mention that its development involved no distillation, meaning it was not trained using another company's AI model. If MAI-Thinking-1 is good, Microsoft clearly doesn't want people to think it's due to OpenAI influence.

Suleyman told The Verge that for Microsoft, “the pivotal moment was the renegotiation of our contract with OpenAI. “That meant we were allowed to train models at larger scales and explicitly pursue superintelligence entirely with our own intellectual property, with our own data, without distillation, with training from scratch.”

Nadella also highlighted Microsoft's recently launched AI cybersecurity tool, MDASH, which he said brings together 100 AI agents to find exploitable bugs "better than any single model." This was clearly a dig at Claude Mythos Preview, which Anthropic introduced in April with much awe and fanfare - and expanded access to just before Build. OpenAI also has its own cybersecurity-focused system, and all three companies will likely use their offerings to position themselves in the government and enterprise markets they desperately need to court.

Microsoft is in a more complex situation with AI agents. The popular open source platform OpenClaw demonstrated a potential path forward for AI agents, and after OpenAI quickly hired its creator, Peter Steinberger, Microsoft (among other companies) is trying to catch up. One of its key strategies is to make OpenClaw work well with Windows. At Build, Nadella said he felt strongly about supporting OpenClaw, and Microsoft employees chatted with developers in the audience about how they were using it.

Steinberger himself made a surprise appearance to a big audience reaction, taking the stage to boast about how OpenClaw had strengthened its security and gained user trust. "What I kept hearing was, 'Peter, I love my Claw, but can I use it at work?'" Steinberger said. “You can now completely run OpenClaw within your enterprise, and we've even made the harness itself a plug-in. » Steinberger said that whether someone trusts Copilot, Codex or another company's coding platform, users can now run OpenClaw on top of it through Windows.

But Microsoft is also promoting its own separate Copilot “super app” that integrates OpenClaw-like agents. A super app is currently a focal point for OpenAI: President Greg Brockman is leading the development of a super app that will connect ChatGPT, the Codex coding platform, and the Atlas web browser. Microsoft's strategy is similar, bringing together a variety of existing Copilot AI assistants. Its agents, called “autopilots,” are designed to act as a useful user interface. Cassidy Williams, senior director of developer advocacy at GitHub, called Copilot "your foundation for development and operations on your computer," demonstrating how multiple agents could perform tasks such as building applications. (In an added effort, Williams demonstrated how she could approve or deny code changes by flashing her computer camera with a thumbs up or thumbs down.)

Autopilots are specifically designed to attract business customers – Nadella called them “autonomous, long-standing, fully corporate compliant agents.” The first that Microsoft will offer is "Scout", billed as "your permanent personal agent", but customers can create and customize their own. Autopilot agents should be able to check an email inbox, join group chats in Teams, view a calendar, and send daily briefings, among other things. As a result, employees on stage at Build repeatedly highlighted Copilot's security tools and safeguards – obviously aimed at calming enterprise customers who may have heard horror stories about tools like OpenClaw.

Suleyman made sure to emphasize, again and again, Microsoft's "humanistic superintelligence" as "AI that prioritizes humanity first" — part of AI companies' recent rebranding of AGI to make it less scary in an era where people are opposing AI more than ever.

Nvidia CEO Jensen Huang, another speaker known for working closely with OpenAI, appeared via video call to tout how Nvidia's RTX Spark chip powers the goals of Microsoft's AI agents. “I could be traveling, I'm on the phone and I can send text messages to my PC... and that would launch the tools on the PC,” Huang said. “The idea that the PC has evolved from a personal computer to a personal AI is really exciting.”

Microsoft spent years betting on OpenAI and, in some ways, that left it behind in the AI race. But as OpenAI and other competitors look to businesses to finally make money, they have clear advantages. Microsoft already has a large customer base and, compared to other AI companies, a reputation for safety and security. And like Google, it also has considerable financial resources, considerable IT resources, and a diversified revenue stream, meaning it can take big bets without taking on much risk.

Suleyman told The Verge: "A lot of people are either chasing startup valuations or about to go public, so we can operate with a little more humility and a little more long-term optimization." He added: “We have the money to be able to buy Anthropic [models] when we need them. We have the option in Azure with 11,000 templates, so people can literally use whatever they want whenever they want, but it gives us time to do it up front.

At the same time, many questions remain unanswered. Microsoft has touted many benchmark wins and breakthroughs for its seven new models, but that doesn't always translate into real-world adoption, and even a new model that advances for a week or two can quickly fall behind. AI super applications are a mostly untested idea. And Microsoft is entering a crowded but still largely disappointing AI agent market with a product we haven't seen in action. There is still plenty of room for his promises to fall flat.