What challenges did ENGIE Digital face before AI implementation?

Anticipate equipment malfunctions and schedule maintenance more effectively while controlling resources and costs. Scale predictive maintenance models across thousands of pieces of equipment and keep pace with changing operational needs.

What was the impact of the AI solution?

More than 1,000 prediction models were developed and trained. Nearly 10,000 pieces of equipment are expected to be connected within 5 years. Estimated savings of €800,000 per year for adopted business units.

Source

Scaled productionEvidence: Medium55/100

ENGIE Digital Uses Amazon SageMaker for Predictive Maintenance at Power Plants

Use case typePredictive maintenanceUpdated Jun 13, 2026

ENGIE Digital built the Robin Analytics and Agathe predictive maintenance platforms for thermal power plants and B2B customer equipment. The platforms use AWS services including Amazon SageMaker, Amazon S3, AWS Glue, and Amazon Athena to develop, train, and deploy predictive maintenance models at scale. The architecture supports a large number of assets and models while emphasizing scalability, security boundaries, and controlled costs.

Organization: ENGIE Digital
Industry: Energy & Utilities
Location: France
Published: June 2026

Reported outcomes

Strategic outcomes

New product / capabilityBuilt predictive maintenance platformsScale & capacityEnabled predictive maintenance at scaleRisk & complianceImproved security compartmentalizationCost efficiencyControlled training and operating costs

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: ENGIE Digital
Provider: AWS
Maturity: Scaled Production
Linked source: AWS Solutions Case Study

The platforms use AWS services including Amazon SageMaker, Amazon S3, AWS Glue, and Amazon Athena to develop, train, and deploy predictive maintenance models at scale

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Predictive Maintenance
2Anomaly Detection
3Asset Reliability

Anticipate equipment malfunctions and schedule maintenance more effectively while controlling resources and costs.
Scale predictive maintenance models across thousands of pieces of equipment and keep pace with changing operational needs.

ENGIE Digital developed the Robin Analytics and Agathe platforms on AWS.
Amazon SageMaker is used to train maintenance models with managed best practices and security compartmentalization.
Amazon S3 and AWS Glue support data storage and transformation, and Amazon EC2 Spot Instances reduce training cost.

Technologies

Amazon SageMaker Amazon S3 AWS Glue Amazon Athena Amazon EC2 Spot Instances

More than 1,000 prediction models were developed and trained.
Nearly 10,000 pieces of equipment are expected to be connected within 5 years.
Estimated savings of €800,000 per year for adopted business units.

Architecture

ENGIE Digital built the Robin Analytics and Agathe predictive maintenance platforms on AWS. Amazon S3 stores the data, AWS Glue supports processing, Amazon Athena is used for analysis, and Amazon SageMaker is used to train maintenance models. The implementation also uses Amazon EC2 Spot Instances to reduce training costs and relies on compartmentalized training tasks for data isolation and security.

Sources & evidence1

Evidence: Medium55/100Evidence strength