We stand at the threshold of a new scientific revolution. The convergence of neuroscience, artificial intelligence and computing has created an unprecedented opportunity to understand intelligence itself. Just as deep-learning architectures inspired by neural circuits have revolutionized AI, insights from machine learning are now transforming our understanding of the brain. This virtuous cycle between biological and artificial intelligence is poised to drive rapid progress in both fields—but only if we can coordinate research at sufficient scale.
Neuroscience has never been better positioned to make transformative discoveries about how intelligence emerges from neural circuits. But our intellectual and financial resources remain fragmented. To truly harness them, we need a new research model that can drive systematic breakthroughs. If we continue to rely on traditional research models that weren’t designed for the scale and complexity of intelligence science, we risk squandering this historic opportunity.
The recent mapping of an entire adult fruit fly brain—a watershed achievement that made headlines worldwide—offers a glimpse of what’s possible. But this breakthrough almost didn’t happen. It required the serendipitous alignment of support from three non-traditional funders: Scientists at the Howard Hughes Medical Institute’s Janelia Research Campus imaged the complete fly brain; the Intelligence Advanced Research Projects Activity drove the development of tools for scalable neural-circuit mapping through its MICrONS program; and the National Institutes of Health BRAIN Initiative provided sustained support for data analysis.
That such a fundamental advance relied on chance alignment reveals a troubling truth: The neuroscience research ecosystem isn’t built to deliver the breakthroughs needed to understand intelligence at scale. Understanding intelligence requires resources and coordination more akin to what goes on in particle physics or astronomy research. The National Institutes of Health’s excellent support for biomedical research isn’t designed for the kind of sustained, large-scale and high-risk investments this challenge demands. Existing private institutions, such as the Allen Institute and the Janelia Research Campus, have made valuable contributions, but they are insufficient for this task. They both have necessarily narrow foci that can’t address the field’s full spectrum of needs. And neither institution is structured to drive rapid translation of insights about intelligence into real-world impact for AI and computing.
The solution isn’t simply better coordination among existing institutions or increased funding through traditional channels. Understanding intelligence demands a fundamentally new approach: a private research institution focused exclusively on intelligence science. To be successful, this institution should combine the proven Advanced Research Projects Agency model of focused, milestone-driven research with the agile capital deployment of modern technology companies. Unlike traditional research organizations, it should maintain its own permanent scientific and engineering staff while also funding and coordinating large-scale initiatives that draw on external teams, such as focused research organizations. The recent success of large-language models shows how concentrated effort and resources can rapidly advance our understanding of intelligence—imagine applying that same focused energy to understanding biological intelligence.
T
he operating model I propose would be a hybrid—part research institute, part funder and part venture studio, bringing together hundreds of researchers across multiple sites equipped with state-of-the-art facilities and computing infrastructure at a scale commensurate with its ambitions. This institution would coordinate multiple large-scale initiatives tackling specific challenges in tool development and large-scale data collection, and its permanent in-house teams would drive data analysis, advances in large-scale simulation and applications to AI and computing. Unlike in traditional research organizations, scientists would be evaluated on impact rather than publication count, with promotion systems rewarding both individual excellence and contributions to team goals. When one group develops breakthrough tools or methods, dedicated teams would ensure rapid deployment across the entire network—creating the kind of information flow common in leading technology companies but rare in academia.Critically, the institution would maintain an internal incubator to spin out promising applications, creating a sustainable funding flywheel that accelerates both basic research and practical advances. During my tenure managing the five-year MICrONS program, our researchers spun out 12 companies that raised $170 million and generated exits of $350 million (so far), dwarfing the federal government’s initial investment. This output demonstrated both the commercial potential of intelligence science and the ability to create sustainable value from ambitious research investments.
The institution’s initial focus should be on transformative tools and datasets that can advance our understanding of intelligence. Beyond connectomics, we need breakthroughs in whole-brain cellular-resolution recordings during complex behaviors, molecularly annotated maps of neuron-glia interactions to understand plasticity, and data-constrained emulation of cognitive function at the whole-brain scale. Each domain requires sustained, high-risk investment at a scale of perhaps $100 million—achievable through a combination of philanthropic endowment and commercial spinoff revenue.
This vision demands investment at substantial scale. The institution would need to launch with at least $1 billion to achieve the critical mass necessary for both breakthrough science and sustainable operations through spinouts. This level of funding may seem ambitious, but it reflects the commercial value potential of intelligence science, which offers a much shorter path from discovery to application than traditional biomedical research. For example, by building on recent advances in optical connectomics, I predict this institution could map the neural circuits underlying decision-making in multiple species in the next five years and then use newly discovered principles of biological computation to rapidly develop language models with improved reasoning capabilities. By commercializing its technology advances, such as more capable AI systems and tools for microscopy and accelerated computing, a well-structured intelligence institute could become self-sustaining within a decade.
There are three potential paths forward. The most promising is the aforementioned privately funded and self-sustaining entity. This route is feasible because private capital is increasingly drawn to brain science, driven by the growing neurotech industry and AI companies seeking to emulate the brain’s information processing capabilities. The alternatives—a new U.S. government agency focused on intelligence science or an international organization resembling CERN (European Organization for Nuclear Research)—face significant political and coordination challenges. No existing funding agency has the necessary focus on understanding intelligence at scale, and building international consensus would likely take decades we don’t have. The private path offers the agility to deploy capital rapidly, the ability to maintain mission focus without political constraints, and the potential for sustainable funding through retained ownership in spinouts.
We are in a unique moment when advances in neuroscience, AI and computing converge. The next great leap in understanding intelligence—both biological and artificial—awaits. Without coordinated action now, we risk spending decades waiting for another chance alignment of funders to produce datasets like the fly connectome, leaving transformative discoveries just out of reach. A private institutional approach represents our best opportunity to build sustainable science at scale—creating an institution that can transform our understanding of intelligence as profoundly as astronomical observatories transformed our understanding of the cosmos. The impact would extend far beyond neuroscience—advancing AI and computing technology while deepening our understanding of human consciousness itself.
