Search and apply for every grant in existence.

This BMFTR funding guideline supports Leitprojekte - flagship, consortium-scale projects - on the use of artificial intelligence across industrial value chains, and it is one of the more substantial German AI calls currently open. The scope is deliberately narrower than general industrial AI: it targets the development of domain-specific AI models and systems, meaning the adaptation or creation of specialised architectures that encode industry-specific vocabulary, process knowledge and physical principles rather than the application of general-purpose foundation models off the shelf. Projects must address three things together. First, vertical anchoring in a real value chain, so the work is demonstrated in the context of an actual industrial sequence rather than a single firm's process. Second, data infrastructure integration with existing industrial IT systems, which is where most industrial AI pilots fail and where BMFTR is explicitly asking for engineering rather than research handwaving. Third, the establishment of open standards for data accessibility and governance across sectors, which signals that BMFTR is funding this as pre-competitive infrastructure and will favour consortia that commit to interoperable, publishable standards over proprietary stacks. Funding goes to consortia combining research institutions and industry. Project sketches (Projektskizzen) are due 1 November 2026, ahead of a second-stage full application for selected consortia - the standard two-stage German process, in which the sketch is the real competition.

The AIMO Interpretability Challenge, part of the NeurIPS 2026 competition program, asks whether interpretability or behavioral methods can tell when a language model genuinely solves a problem and when its reasoning is spurious. Given an unseen olympiad math problem and an AIMO 3 model, a system must label the model's solution as robust or spurious, and entries are ranked by accuracy. Methods can use white-box signals such as weights, activations and chain-of-thought, or black-box analysis. There are two tracks: a Main Track with no model size limit, and a Small Models Track for models under 10B parameters. The models include Qwen3.5-4B, Olmo-3-7B-Think, DeepSeek-R1-0528-Qwen3-8B and gpt-oss-120b. The $12,500 prize pool pays $2,500, $1,000 and $500 in the Main Track and $1,500, $700 and $300 in the Small Models Track. Each track also has a $1,000 White-Box Award, and $4,000 goes to writeup prizes. Submissions are offline Codabench bundles with a 3,600-second limit. Final submissions are due 1 November 2026, technical reports by 15 November, and the workshop is on 12 December 2026 in Paris.

The CHIPS Incentives Program aims to catalyze long-term economically sustainable growth in the domestic semiconductor industry in support of U.S. economic and national security. This is the second Notice of Funding Opportunity under this program and seeks applications for projects for the construction, expansion, or modernization of commercial facilities for semiconductor materials and manufacturing equipment. For more information, additional resources, and instructions on how to apply, please visit chips.gov. Note that all applications must be submitted through the CHIPS Incentives Portal at https://applications.chips.gov. If you have any questions on how to apply, please email apply@chips.gov. Funding Opportunity Number: 2023-NIST-CHIPS-SMME-01. Assistance Listing: 11.037. Funding Instrument: CA. Category: ST.