Can AWS MAP fund an AI project?
Summary
MAP is a migration program. It funds the movement of existing workloads to AWS, not the development of new applications. A net-new AI project being built on AWS does not qualify for MAP regardless of which services it uses. That said, AI workloads can qualify for MAP when they involve migrating existing infrastructure — training pipelines, data environments, or inference workloads moving from on-premises or another cloud to AWS. For net-new AI builds, AWS has a separate mechanism: Proof of Concept (PoC) funding. The two programs are not mutually exclusive, and many AI-focused projects qualify for both depending on what is being built versus what is being moved.
This is one of the most common points of confusion around AWS MAP funding. Here’s how this breaks down.
MAP is for migration, not development
MAP funds the movement of workloads that already exist somewhere outside AWS. The AI nature of a workload does not change that eligibility logic. What matters is whether something is moving from outside AWS to AWS.
A company migrating AI training pipelines from on-premises GPU infrastructure to AWS would have those workloads evaluated under the same MAP criteria as any other migration. A company migrating model inference infrastructure from Azure or GCP to AWS would qualify the same way. The workloads are AI workloads, but the program logic applies because they are migrating.
A net-new AI project, like a new agent, a new Bedrock-powered application, or a new document processing pipeline built from scratch, does not qualify for MAP. Nothing is moving. There is no migration.
For net-new AI builds: PoC funding
AWS has a separate funding mechanism for new AI development: Proof of Concept (PoC) funding. AWS PoC credits offset the AWS service costs of building a defined prototype. Projects using Amazon Bedrock or SageMaker as the primary workload are prioritized.
PoC funding covers AWS compute and API costs. It does not cover consulting time or software licenses. Eligibility requires a defined technical scope, measurable success criteria, and a credible path to production. The credit amount depends on scope and is determined through the application process.
Both programs can apply to the same project
A company migrating AI infrastructure to AWS may qualify for MAP on the migration itself. At the same time, a new Bedrock-based application being built alongside that migration may qualify for separate PoC funding. The two programs address different parts of the same project.
Tech 42 works with both programs. We help customers identify which workloads qualify for MAP and which AI builds are candidates for PoC funding, and we submit the appropriate applications through our AWS partner relationship. Our AI Infrastructure Migration service and AgentCore Accelerate Program have both been funded through AWS in this way.
