As cloud adoption accelerates, estimating deployment costs early in the system design phase becomes critical for informed architectural decisions. However, existing approaches often require detailed configurations or lack integration with high-level design models, making early cost prediction challenging. This paper addresses this gap by introducing a Data Flow Diagram (DFD)-based approach for early-stage cost estimation in distributed systems. Our method enhances traditional DFDs with structured annotations and models system interactions as attribute-rich sentences to estimate cloud resource consumption. To demonstrate the practicality of the approach, we applied it to a microservice-based ticket reservation system as a use case. Validation results show that our model achieves an average accuracy of 96.69% compared to actual deployment costs and outperforms Infracost tool by 11.22%. These findings highlight the model’s potential to bring cost-awareness into the design phase, enabling more sustainable and financially optimized cloud architectures.