Microsoft Builds Independent Superintelligence Team, Signaling New Era for Global AGI Race

Microsoft creates a “Superintelligence” division

Microsoft has established an internal division called “Superintelligence”, led by former Allen Institute for AI (AI2) chief executive Ali Farhadi. Core members are recruited from AI2 and the University of Washington.

Goal and strategic purpose

The division’s explicit target is artificial general intelligence (AGI). Rather than improving specific product features, the team is tasked with exploring the fundamental theory and architecture of AGI, thereby giving Microsoft a native capability that can operate independently of external partners.

This structure functions as a strategic redundancy and an autonomous evolution path: it does not aim to replace the existing partnership with OpenAI immediately, but ensures that Microsoft retains its own “engine” for AGI research if external collaborations change.

Risk of dependence on OpenAI

Microsoft’s collaboration with OpenAI has delivered large investments, exclusive cloud services, and deep product integration, propelling ChatGPT to global prominence. However, OpenAI remains an independent company whose research direction, technical decisions, and governance are outside Microsoft’s direct control. For a company seeking to dominate the next computing era, relying on an external entity for the core AGI effort introduces a clear strategic risk.

Why Ali Farhadi?

Farhadi comes from the Allen Institute for AI, a nonprofit, open‑research organization founded by Paul Allen. His background emphasizes long‑term, academic research on core cognitive abilities—vision, reasoning, and common‑sense understanding—that current large language models struggle to master.

Integrating this “pure‑intelligence” expertise into Microsoft’s industrial R&D signals a shift from merely scaling existing large models toward combining deep scholarly insight with massive compute and data resources.

“True general intelligence requires understanding the world rather than merely describing it. This demands fundamental innovation in model architecture and training paradigms.”

Changing competitive landscape

Earlier AGI competition centered on pure AI players such as OpenAI, Google DeepMind, and Anthropic. Microsoft, previously viewed as an “amplifier” for OpenAI, is now entering the arena directly, aiming to develop the “strongest warrior” itself. The battlefield is evolving into a mixed clash between technology giants and AI‑native companies.

For rivals like Google, Meta, and Amazon, the rapid absorption of top AI talent into an ambitious industrial lab serves as a warning that future breakthroughs may emerge from entities that combine visionary research with virtually unlimited resources.

Key challenges include preserving startup‑like agility within a large corporate structure and balancing short‑term product pressures with high‑risk, long‑term fundamental research.

Strategic implications

The “Superintelligence” team gives Microsoft unprecedented flexibility: it can continue collaborating with OpenAI while maintaining an internally developed AGI capability as a fallback, thereby securing strategic autonomy and the ability to influence the next AI paradigm.