Microsoft hires former Ai2 CEO and top researchers for AI advancement
Microsoft is bringing Ali Farhadi, former CEO of the Allen Institute for AI, and three other senior researchers from the Seattle-based nonprofit to lead model development work within the company's Superintelligence team.
Farhadi, along with researchers Hanna Hajishirzi, Ranjay Krishna, and Sophie Lebrecht (former Ai2 chief operating officer), will join Mustafa Suleyman's organization while retaining faculty positions at the University of Washington's Allen School of Computer Science and Engineering. Farhadi departed Ai2 on March 12 after leading the institute for more than two and a half years.
Why Microsoft pursued these hires
The moves reflect Microsoft's effort to reduce reliance on OpenAI for frontier AI models while competing against Amazon and Google. Suleyman's Superintelligence team, formed in November, focuses on developing advanced foundation models with billions in compute backing.
Ai2 has built particular strength in open-source model development and training efficiency-areas where the nonprofit has outperformed much larger organizations. Hajishirzi co-leads the OLMo open-source language model project and serves as co-principal investigator on a $152 million, five-year National Science Foundation and Nvidia initiative for scientific AI models. Krishna developed Ai2's Molmo multimodal models.
Farhadi is a computer vision specialist who co-founded Xnor.ai, which Apple acquired in 2020 for an estimated $200 million. He later led machine learning at Apple before returning to Ai2 as CEO.
Shifting funding priorities at Ai2
The departures reflect changing funding dynamics at the nonprofit. Ai2 was historically backed by Paul Allen's Vulcan Inc. and his estate, but the Fund for Science and Technology-a $3.1 billion foundation created under Allen's instructions-is now the primary backer.
The fund, led by CEO Dr. Lynda Stuart, favors applied AI uses over the expensive work of building frontier models. While all Ai2 programs for 2026 remain fully funded, the foundation is shifting from annual grants to proposal-based funding. Future support is expected to prioritize real-world applications rather than open-source foundation model development.
This funding shift explains why researchers focused on model development chose to move to Microsoft, where resources for that work are substantially larger.
Ai2's response
Interim CEO Peter Clark acknowledged the departures in a statement, saying Ai2 remains committed to its mission and partnerships with the NSF and Nvidia. "These initiatives are backed by a broad, experienced team with the expertise and continuity needed to carry this work forward," Clark said.
A Fund for Science and Technology spokesperson said Ai2's work and mission remain unchanged, noting that the foundation's broader program strategies are still under development.
Building Microsoft's research bench
Microsoft has already hired researchers from Google DeepMind, Meta, OpenAI, and Anthropic. The addition of the Ai2 and UW group strengthens the company's roster of scientists working on foundation models and training approaches.
In a LinkedIn post about the hires, Suleyman praised Farhadi for leading Ai2 to release more than 100 models in a single year and called Hajishirzi "one of the most cited researchers of natural language processing in the world." He described the group as helping Microsoft pursue "humanist superintelligence: safer, controllable, more capable AI systems in service of humanity and our toughest problems."
Professionals in AI research may benefit from exploring AI Research Courses and Generative AI and LLM Courses to deepen expertise in foundation model development and training methodologies.
Your membership also unlocks:
