Astera Neuro & AGI

Astera Neuro & AGI is the flagship research division of the Astera Institute, a 501(c)(3) private foundation founded in 2020 and headquartered in Berkeley, California. The institute was founded by Jed McCaleb, the co-founder of Ripple and the Stellar Development Foundation, who serves as Co-Founder and CEO with his focus centered on this program. The Neuro & AGI program is built on the premise that AI will be the most transformative technology in human history, and that understanding how biological intelligence works is the most promising path to developing safe and beneficial AGI. The program departs from the dominant paradigm of simply scaling compute and data, instead investigating how natural systems achieve sophisticated cognition on the power consumption of a lightbulb. The program comprises three interconnected research efforts. The Neuroscience effort, launched in December 2025 under Chief Scientist Doris Tsao (a former UC Berkeley professor and Howard Hughes Medical Institute Investigator), aims to decode how the brain transforms sensory inputs into conscious experience, using the hypothesis that the brain employs a compositional architecture combining elemental units through neural syntax. The Neuro-AI effort, led by Head of AI Dileep George (who joined in February 2026 from Google DeepMind, where he worked on memory, planning, and structure learning, and who previously co-founded Numenta and Vicarious), builds AI systems that learn like biological systems, tackling challenges such as long-range planning, continuous learning from limited data, and robust generalization. Research Lead Miguel Lazaro-Gredilla (also formerly Google DeepMind) spearheads work on world models using hierarchical latent variables for long-horizon planning. The AI Safety effort runs a program called Simplex, which develops geometric frameworks to uncover the internal structures that produce model capabilities, with the goal of building interpretability tools that can scale to frontier systems. Astera operates with over $1 billion in committed resources for this program over the coming decade. The institute held $2.62 billion in total assets as of fiscal year 2024. The team operates with access to significant computational resources and is unconstrained by typical funding, profit, or publication pressures. As of early 2026, Astera is actively recruiting researchers and engineers with machine learning expertise and neuroscience interest, and accepts residency applications.