NextFin News - Microsoft has appointed Ali Farhadi, the former chief executive of the Allen Institute for Artificial Intelligence (Ai2), as a corporate vice president within its rapidly expanding AI division. The move, confirmed on March 23, 2026, places Farhadi directly under Mustafa Suleyman, who was recently repositioned by U.S. President Trump’s era of tech consolidation to lead Microsoft’s "Superintelligence" efforts. Farhadi’s arrival follows his sudden departure from the non-profit Ai2 earlier this month and signals a decisive shift in Microsoft’s strategy toward building sovereign, high-efficiency frontier models that reduce its long-standing dependency on OpenAI.
The hiring of Farhadi is a calculated talent grab that brings deep expertise in "on-device" AI and open-source research to Redmond. Before leading Ai2, Farhadi co-founded Xnor.ai, a startup specializing in running complex AI models on low-power hardware, which Apple acquired for $200 million in 2020. By bringing Farhadi into the fold, Suleyman is assembling a "super-team" designed to solve the industry’s most pressing bottleneck: the massive computational and energy costs of generative AI. While Microsoft’s partnership with OpenAI remains its most visible asset, the internal reorganization around Suleyman and Farhadi suggests the software giant is preparing for a future where it owns the underlying "frontier models" rather than just renting them.
This internal pivot comes at a delicate moment for Microsoft’s consumer AI ambitions. Internal data suggests that Copilot, the company’s flagship AI assistant, has reached only about 3% of Microsoft’s enterprise user base, with daily active users hovering around 6 million. This adoption curve is significantly shallower than the viral growth seen by ChatGPT, prompting a strategic reset. Suleyman, who joined Microsoft in 2024 after co-founding DeepMind and Inflection AI, was recently tasked with focusing exclusively on "Superintelligence" and world-class model delivery over a five-year horizon. Farhadi’s role will likely involve bridging the gap between these massive "frontier" models and the efficient, localized execution required for Windows and mobile devices.
The competitive landscape is also forcing Microsoft’s hand. With Amazon and Google aggressively verticalizing their AI stacks—from custom silicon to proprietary models—Microsoft cannot afford to be a mere distribution layer for third-party technology. The "Superintelligence" team is Microsoft’s bid for self-sufficiency. By recruiting a leader like Farhadi, who has spent years navigating the intersection of academic rigor at Ai2 and commercial efficiency at Xnor.ai, Microsoft is betting that the next phase of the AI race will be won by those who can make "superintelligence" both portable and private.
For the broader industry, Farhadi’s move from a non-profit research bastion to a corporate titan underscores the intensifying "brain drain" from open science into private labs. Ai2 has long been a defender of open-source AI, but as the capital requirements for training state-of-the-art models soar into the billions, even the most prestigious non-profits are finding it difficult to keep pace. Microsoft now possesses one of the most formidable leadership benches in the history of computing, yet the pressure to convert this intellectual capital into "world-class models" that justify its trillion-dollar valuation has never been higher.
Explore more exclusive insights at nextfin.ai.
