Microsoft's AI chief, Mustafa Suleyman, emphasized a foundational commitment, stating, “We have to prove that we can do everything that we need to from the ground up.”
During Microsoft’s annual Build conference on Tuesday, the company unveiled a comprehensive suite of new and expanded AI initiatives. These included a proprietary super app, advanced in-house reasoning models, a specialized cybersecurity tool, and AI agents inspired by the OpenClaw paradigm. This extensive announcement underscored a clear message: Microsoft is asserting its position as a dominant force in the AI landscape, finally demonstrating its full capabilities.
For several years, Microsoft's AI strategy heavily relied on its foundational and exclusive collaboration with OpenAI. However, this partnership, marked by internal complexities, transitioned into a less formal arrangement, with an effective separation occurring in late April, though Microsoft continues as OpenAI’s primary cloud partner for the foreseeable future. This year’s Build conference exuded an atmosphere of renewed independence, with Microsoft CEO Satya Nadella remarking onstage, “It’s always fun to be at developer conferences in times of great change,” adding that such events are about “coming to grips with the new opportunity.”
Mustafa Suleyman, in an interview with The Verge, articulated Microsoft's ambition even more directly.
“The goal is to prove that we can become one of the top four labs in the world,” Suleyman declared. He identified the current leading labs as Google DeepMind, OpenAI, and Anthropic, acknowledging, “We are not one of them at the moment, and that’s always been my intention. It’s why I came here. I want to build the very best frontier models in the world, fully multimodal, and in order to do that, we have to prove that we can do everything that we need to from the ground up, and we’re not just going to take from others.”
A key strategic move at Build involved Microsoft's efforts to advance its own AI model development. Suleyman introduced MAI-Thinking-1, the company's inaugural reasoning model, alongside six other new models specialized in image processing, voice recognition, transcription, and coding. Microsoft described the medium-sized MAI-Thinking-1 model, primarily aimed at enterprise clients, as “built from scratch for serious math, coding, and real-world enterprise deployment.” While Microsoft trails OpenAI and Anthropic in this domain—OpenAI having released reasoning models in late 2024—Suleyman highlighted MAI-Thinking-1's strong performance on coding benchmarks and its competitive pricing, noting it was more cost-effective than some OpenAI alternatives, a significant advantage amidst current AI expenditure concerns.
Despite years of collaboration and potential insights gained from OpenAI, Suleyman explicitly stated that MAI-Thinking-1’s development involved no distillation, meaning it was not trained using another company's AI model. This clarification underscores Microsoft's intent to establish its proprietary model's quality independently, rather than attributing it to OpenAI's influence.
Suleyman further explained to The Verge that a “pivotal moment was renegotiating our contract with OpenAI. That meant that we were allowed to train models at a larger scale and explicitly pursue superintelligence entirely with our own IP, with our own data, no distillation, training from scratch.”
Nadella also showcased MDASH, Microsoft’s recently launched AI cybersecurity tool, which he stated leverages 100 AI agents to identify exploitable vulnerabilities “better than any single model.” This announcement appeared to be a subtle competitive response to Anthropic’s Claude Mythos Preview, which garnered considerable attention and expanded access just prior to Build. OpenAI also maintains its own cybersecurity-focused system, indicating an emerging competition among these major players for dominance in critical government and enterprise markets.
Microsoft faces a more intricate landscape concerning AI agents. The popular open-source platform OpenClaw demonstrated significant potential in this area, and after OpenAI swiftly recruited its creator, Peter Steinberger, Microsoft, among others, is striving to catch up. A core part of Microsoft's strategy is to ensure seamless integration of OpenClaw with Windows. Nadella reiterated his strong commitment to OpenClaw support at Build, with Microsoft employees actively engaging developers in the audience about their usage experiences.
In a notable moment, Steinberger made a surprise appearance, met with enthusiastic audience reception. He highlighted OpenClaw’s enhanced security and growing user trust. “What I kept hearing was, ‘Peter, I love my Claw, but can I use it at work?’” Steinberger recounted. He confirmed, “You can totally run OpenClaw inside your company now, and we even made the harness itself a plug-in.” Steinberger elaborated that users could now operate OpenClaw via Windows, irrespective of whether they relied on Copilot, Codex, or another company's coding platform.
Concurrently, Microsoft is advancing its distinct Copilot “super app,” designed to integrate OpenClaw-esque agents. The concept of a super app is a key focus for OpenAI, with its president, Greg Brockman, leading development for a platform that will unify ChatGPT, the Codex coding platform, and the Atlas web browser. Microsoft's approach mirrors this, consolidating various existing Copilot AI assistants. Its specialized agents, termed “Autopilots,” are engineered to provide an intuitive user interface. Cassidy Williams, GitHub’s senior director of developer advocacy, characterized Copilot as “your home base for development and operations on your computer,” illustrating how multiple agents can collaboratively execute tasks such as application building. As an added demonstration, Williams showed how she could approve or reject code changes simply by gesturing with a thumbs-up or thumbs-down to her computer camera.
These Autopilots are specifically tailored for business clientele, with Nadella describing them as “autonomous, long-running agents with full enterprise compliance.” The first Autopilot Microsoft will introduce is “Scout,” positioned as “your always-on personal agent,” though clients retain the flexibility to build and customize their own. These agents are designed to manage tasks such as reviewing email inboxes, participating in Teams group chats, checking calendars, and dispatching daily briefings. Consequently, Microsoft employees at Build consistently emphasized Copilot’s robust security tools and guardrails, aiming to reassure enterprise clients who may have concerns stemming from reports about other AI tools like OpenClaw.
Suleyman repeatedly underscored Microsoft’s vision for “humanist superintelligence,” defining it as “AI that prioritizes humanity first.” This framing aligns with a broader industry trend among AI companies to rebrand Artificial General Intelligence (AGI) in a less intimidating manner, particularly as public sentiment increasingly questions AI's societal impact.
Jensen Huang, CEO of Nvidia and a known collaborator with OpenAI, made a virtual appearance to highlight how Nvidia’s RTX Spark chip is instrumental in powering Microsoft’s AI agent ambitions. Huang shared an illustrative scenario: “I could be traveling and I’m on the phone and I can text my PC … and it would fire up the tools on the PC.” He concluded, “The idea that the PC evolved from a personal computer to a personal AI is just really exciting.”
While Microsoft's long-standing reliance on OpenAI may have caused some initial delays in its independent AI progress, it now possesses significant advantages as competitors like OpenAI increasingly target the enterprise sector for monetization. Microsoft boasts a substantial existing client base and a well-established reputation for safety and security, setting it apart from many newer AI companies. Furthermore, akin to Google, Microsoft commands vast financial resources, extensive computing infrastructure, and a diversified revenue portfolio, enabling it to undertake ambitious AI investments with mitigated risk.
Suleyman informed The Verge, “There’s a lot of people who are either like chasing startup valuations or about to IPO, so we can operate with a little bit more humility and a little bit more long-term optimization.” He elaborated, “We’ve got the money to be able to buy Anthropic [models] when we need to. We’ve got the optionality in Azure with 11,000 models, so people can use literally whatever they want whenever they want, but that buys us the time to do it right from the start.”
Nevertheless, several critical questions remain unanswered. While Microsoft presented numerous benchmark successes and advancements for its seven new models, such results do not always translate into widespread real-world adoption, and even temporarily leading models can quickly be surpassed. The concept of AI super apps is largely unproven. Moreover, Microsoft is entering an AI agent marketplace that is already crowded yet still largely underperforming, with its own product yet to be showcased in live action. Consequently, there remains considerable potential for these ambitious promises to fall short of expectations.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.