Microsoft’s Superintelligence Quest & The New Race Against OpenAI

In a strategic pivot that signals a new phase in the global AI arms race, Microsoft is embarking on a quest to build its own “superintelligence.” The tech giant, which has a market capitalization of over $3 trillion, has long been viewed as being primarily dependent on its partner, OpenAI. However, under the leadership of its new AI CEO, Mustafa Suleyman, Microsoft is now building a “world-class, frontier-grade research capability in-house” to become “self-sufficient in AI.”indianexpress+3​

This dramatic shift comes just weeks after Microsoft and OpenAI fundamentally restructured their partnership agreement. The revised terms now grant Microsoft the freedom to pursue Artificial General Intelligence (AGI) independently. This move sets the stage for a new era of intense competition, not just with rivals like Google and Meta, but potentially with its own closest ally, OpenAI. Suleyman’s newly formed “superintelligence team” is tasked with a monumental goal: to create a new class of AI that is not only powerful but also safe, controllable, and explicitly designed to serve human interests.businessinsider+5​

Personal Experience: “For the past two years, the narrative has been simple: Microsoft provides the capital and cloud infrastructure, and OpenAI provides the frontier model innovation. This created a powerful but ultimately dependent dynamic for Microsoft. The formation of Suleyman’s superintelligence team, combined with the renegotiated OpenAI deal, is the most significant strategic shift we’ve seen from Microsoft in the AI era. It’s a clear declaration that while the partnership with OpenAI remains, Microsoft is no longer content to simply be the power source; it intends to build its own engine.”

This BroadChannel analysis breaks down what this strategic pivot means for the AI industry, the specifics of the new Microsoft-OpenAI deal, and Suleyman’s vision for a “humanist superintelligence.”

The Mission: Self-Sufficiency and Humanist AI

Mustafa Suleyman’s vision for Microsoft AI is clear and ambitious. He has stated that the company “needs to be self-sufficient in AI” and to achieve this, it must “train frontier models of all scales with our own data and compute at the state-of-the-art level.” This marks a departure from Microsoft’s previous strategy, which heavily focused on leveraging and post-training models developed by OpenAI.indianexpress+1​

The new superintelligence team has two primary objectives:

1. Achieve Frontier Research Capability: The team’s immediate remit is to build a world-class research unit capable of solving the fundamental problems that are currently barriers to creating more advanced AI. This includes tackling challenges like continual learning (adding new knowledge to existing models without full retraining) and transfer learning (allowing one model to teach another).businessinsider​
2. Build “Humanist Superintelligence”: Suleyman has repeatedly emphasized that the goal is not to create an “unbounded and unlimited entity with high degrees of autonomy.” Instead, the focus is on “humanist superintelligence”—AI that is “carefully calibrated, contextualised, and within limits,” with the ultimate purpose of keeping humans “at the top of the food chain.” This involves a deep focus on AI safety and building controllable systems.timesofindia.indiatimes+2​

The practical applications for this humanist AI are vast, with Suleyman citing goals like revolutionizing healthcare with expert-level medical diagnostics, accelerating breakthroughs in clean energy, and ultimately “reducing the cost of living for billions of people over the next 10 years.”cognativ+1​

The New Deal: How Microsoft Unshackled Itself from OpenAI

This new quest for self-sufficiency was made possible by a fundamental restructuring of the Microsoft-OpenAI partnership agreement. While the two companies remain close partners, the revised terms, finalized in October 2025, introduce a new level of flexibility and independence for both parties.vktr+1​

Key Changes in the Agreement:

Provision	New Term	Implication for Microsoft
Independent AGI Pursuit	Microsoft is now explicitly permitted to pursue AGI independently or with other partners.	Microsoft is no longer solely reliant on OpenAI to achieve its AGI ambitions and can build its own frontier models. pulse2​
Third-Party Partnerships	OpenAI can now co-develop products with third parties, though any API products must remain exclusive to Azure.	Microsoft benefits from a wider range of AI products being available on its cloud platform, but OpenAI is no longer exclusive to them. pulse2​
Extended IP Rights	Microsoft’s IP rights to models and products developed by OpenAI are extended through 2032 and now include post-AGI models.	This ensures Microsoft has long-term access to and benefits from OpenAI’s innovations, even after AGI is achieved. pulse2​
AGI Verification	The declaration of AGI will now be verified by an independent expert panel, not just by OpenAI’s board.	This introduces a more objective and transparent process for determining when AGI has been reached, a key trigger in the partnership agreement. pulse2+1​

This new agreement effectively transforms the relationship from one of dependence to one of strategic alignment. Microsoft maintains its deep investment and access to OpenAI’s technology while gaining the freedom to forge its own path in the race to superintelligence.pureai​

The Competitive Landscape: A Five-Way Race

The formation of Microsoft’s superintelligence team officially places it in a five-way race for AGI, competing directly with:

• Google: With its deep research roots in AI and its powerful Gemini models, Google is a formidable competitor.
• Meta: By open-sourcing its powerful Llama models, Meta is pursuing a strategy of commoditizing the AI market.
• Anthropic: Focused on AI safety with its “Constitutional AI” approach, Anthropic competes for enterprise customers who prioritize caution.
• OpenAI: Microsoft’s closest partner is now also its most direct competitor in the quest to build the world’s most advanced AI.businessinsider​

Suleyman has acknowledged this competitive landscape, stating that Microsoft will not be “religious” about which models it uses. While the company will develop its own in-house “MAI” models, it will continue to offer and integrate the best available technology, whether from OpenAI, Anthropic, or the open-source community, to ensure its products are “AI-first.”businessinsider​

The Humanist Imperative: AI Safety as the North Star

Throughout his public statements, Suleyman has made it clear that safety is the “number one most important thing that humanity focuses on.” He has warned of the risk of AI systems that “get extremely smart and run away from us” and has emphasized that the design of these systems must be rooted in a “humanist intent.”indiatoday+2​

This philosophy rejects the idea of building AI that simulates consciousness or triggers human empathy. Instead, the goal is to build powerful tools that are explicitly designed to be controllable, interpretable, and to serve human needs. This cautious and deliberate approach may prove to be more expensive or less efficient in the short term, but Suleyman and Microsoft are betting that it is the only sustainable path to building truly beneficial superintelligence.economictimes+3​

As Microsoft embarks on this ambitious new chapter, the AI world will be watching closely. The race to superintelligence is no longer just about who can build the most powerful model; it’s about who can build the most powerful model that remains firmly in service of humanity.