Unconventional AI Academic Research Fund for Novel Computing Paradigms in Generative AI

The Unconventional Academic Research Grant is a small, sharply targeted fund - USD 500,000 total, five awards of USD 100,000 each - aimed at a question most generative AI funding ignores: whether the digital architectures currently used to run large models are the right ones at all. Six focus areas define the scope and they are unusually specific. Unconventional Circuits covers analog and digital circuits functioning as physical neural networks. Systems Architecture covers frameworks for executing neural networks across heterogeneous computing cores. Neural Network Dynamics covers architectures that exploit analog system dynamics for efficiency advantages. Neural Network Data Movement is the most concretely stated - architectures enabling more than 99 percent of computation through recurrence without accuracy loss. AI Theory covers theoretical and dynamical systems research informing unconventional architecture design. 3D Integration covers computing and memory integration with manufacturing scalability achievable within five years. That five-year manufacturability constraint recurs across the programme and is the clearest signal of what the funder wants: physically realisable alternatives, not thought experiments. The application process is two-stage and deliberately low-burden at entry. A pre-proposal requires a project abstract of at most 500 words, a nominated focus area, team information of at most 200 words, and - distinctively - a description of the applicant's most unconventional published work in at most 300 words, which is effectively the screening question. Invited full proposals run to five technical pages and must include rationale, three potential failure points and expected impact. Requiring applicants to name their own failure modes is a meaningful filter. The 2026 cycle has closed, with pre-proposals due 15 May 2026 and full proposals 28 July 2026; researchers in this area should watch for the next round.

The MicroChat CS & AI Literacy Grant provides free site licensing, professional development materials, and live training to equip K-12 educators with hands-on AI and computer science literacy tools. Awards (5-50 MicroChat licenses valued at $200 each) enable a train-the-trainer model in which coaches and specialists reach hundreds of teachers and thousands of students with AI literacy instruction.

The OpenAI Researcher Access Program provides up to $1,000 in OpenAI API credits per award to support academic and independent researchers studying AI capabilities, safety, alignment, and societal impact using frontier generative models. Applications are reviewed on a rolling quarterly basis (March, June, September, and December), lowering the barrier for researchers to experiment with large language models and multimodal AI systems.