With the Amazon Research Awards Amazon supports scientific endeavors that develop new methods and applications in the field of artificial intelligence and related technologies.

The current call for proposals includes nine research focal points:

1. Agentic AI: Development of autonomous AI agents, multi-agent systems, and new forms of collaboration between humans and AI.

2. AI for Information Security: AI-based approaches to improve information security.

3. Amazon Security: Research on security-related challenges and AI-powered security solutions.

4. Automated Reasoning: Mathematical methods for verifying and securing complex computer systems.

5. AWS Cryptography: New approaches to cryptographic procedures and secure digital infrastructures.

6. Build on Trainium: AI-powered optimization of computing cores and software frameworks for AWS Trainium hardware.

7. Multilingual and Contextual AI Evaluation: Methods for evaluating AI systems across different languages and usage contexts.

8. On-Device Multimodal Perception: Multimodal AI perception directly on end devices.

9. Synthetic Data Generation for Multimodal Perception: Generation of synthetic training data for multimodal perception systems.

The call is aimed at both theoretical research approaches and practical developments and novel technical procedures.

Agentic AI and Security as Research Focuses

A special focus is on the further development of agentic AI systems, which can independently plan and execute complex tasks.

Research approaches for the collaboration of multiple AI agents, the automation of workflows, and the use of agentic systems in scientific research are particularly sought. Reliable verification of autonomous AI activities as well as questions of security and responsible usage are also relevant topics.

Further calls for proposals deal with automated logical reasoning, information security, and cryptography. This includes, for example, the mathematical verification of software and AI systems, as well as methods for detecting and preventing security risks.

More Efficient AI Hardware and Better Model Evaluation

In addition to developing new AI methods, Amazon promotes research on the technical infrastructure of powerful AI systems. In the topic area Build on Trainium , approaches are central, with which AI agents can support the optimization of computing cores and software frameworks. Further research fields deal with the evaluation of multilingual AI systems and the processing of different data modalities.

In multimodal perception, for example, it is about efficiently processing visual and other sensory information. Synthetic data should offer new possibilities to train and evaluate such systems. Thus, the call covers both fundamental AI research and questions of practical implementation and computational efficiency.

Financial Resources and AWS Credits for Research Projects

The Amazon Research Awards offer financial support, AWS credits, or a combination of both, depending on the call for proposals. The specific funding amount differs among the nine calls. For example, the call for agentic AI provides an average of up to USD 70,000 in financial resources and up to USD 50,000 in AWS credits. The final approval is determined individually.

The financial resources are granted as a one-time, non-dedicated endowment to the respective university or research institution. Amazon does not acquire any rights to the intellectual property of the resulting research outcomes through the funding. The publication of scientific results and the provision of open-source code are expressly welcomed.

Funded researchers also gain access to professional contacts at Amazon as well as training opportunities and exchange formats with the company's scientists.

Who Can Apply?

The call is aimed at full-time academic staff of accredited universities and permanently employed researchers at recognized non-profit non-governmental organizations.

Scientists at relevant institutions in Germany can also submit applications. Only one application per responsible research leader (Principal Investigator) is allowed per call for proposals.

Further participation and exclusion criteria apply, such as for certain employment relationships, conflicts of interest, and previous work for Amazon. Interested parties should therefore check the respective calls for proposals and official program rules before submitting an application.

Application Deadline

Research applications can be submitted until November 5, 2026, at 08:59 AM (CET).

The application phase has been running since October 1, 2026. Applications are evaluated particularly based on scientific quality, originality, and the potential of their results. The topic-specific calls for proposals contain additional requirements for the project description and other required information.

For universities and research institutions in the Bavarian AI ecosystem, the program offers an additional opportunity to support research projects on agentic AI, security, hardware optimization, and multimodal AI systems financially and technically.