Serverless Functions

Run code without managing servers. Pay only when your function executes.

What each cloud calls it

AWS Lambda, Azure Functions, and GCP Cloud Functions are mature platforms with deep ecosystem integration: event sources, dead-letter queues, provisioned concurrency, and fine-grained IAM. Scaleway Serverless Functions is the only European managed equivalent. OVHcloud, STACKIT, IONOS, and Hetzner do not offer a serverless functions product at the time of writing.
Scaleway Serverless Functions supports Python, Node.js, Go, PHP, and Rust runtimes. Cold start behaviour is comparable to Lambda at similar memory allocations. There is no provisioned concurrency equivalent, so latency-sensitive workloads that cannot tolerate occasional cold starts need a different strategy.
Trigger sources on Scaleway are limited compared to Lambda. Direct HTTP invocation and Scaleway-internal event sources (Queues, Cron) are supported. Lambda's breadth of event sources (S3, DynamoDB Streams, Kinesis, SNS, SQS, EventBridge) has no equivalent and likely will not have one in the near term.
Maximum execution duration and memory limits differ. Lambda allows up to 15 minutes and 10 GB memory. Scaleway Serverless Functions are designed for short-lived, lightweight workloads. Verify your function's timeout and memory requirements before assuming a direct swap.

Scaleway bills per request and per GB-second of execution, with a free tier. Pricing is broadly competitive with Lambda at low-to-medium invocation volumes. At very high request rates, the cost difference narrows.
Lambda pricing varies by region, runtime, and whether you use ARM64 (Graviton). ARM64 Lambda functions cost roughly 20% less per GB-second than x86. Scaleway does not yet offer an ARM64 serverless runtime.
Neither platform charges for idle time, which is the main cost advantage of serverless over always-on compute. This comparison holds across both providers.

Serverless Framework and AWS SAM are AWS-centric and have limited or no Scaleway support. For Scaleway, Terraform is the primary IaC tool; the official Scaleway provider covers function deployment, namespace management, and trigger configuration.
The Scaleway CLI (scw) can deploy and invoke functions directly. It is less feature-rich than the AWS CLI for Lambda operations but covers the common deployment workflow.
Observability is more limited on Scaleway. Lambda integrates natively with CloudWatch for logs, metrics, and X-Ray tracing. Scaleway Cockpit provides basic metrics and log shipping, but distributed tracing requires additional instrumentation.
For workloads that need serverless-style scaling but more trigger flexibility, Scaleway Serverless Containers (see Managed Containers) is worth evaluating alongside Serverless Functions.

Scaleway is ISO 27001, SOC 2, and SecNumCloud certified. Serverless function code and execution happen inside French or Dutch data centres depending on region selection. No data transits US-jurisdiction infrastructure.
Lambda in EU regions (Frankfurt, Ireland, Stockholm, Paris) carries ISO 27001 and SOC 2, but invocation metadata and control-plane operations route through US-based AWS systems. This is the core sovereignty concern for regulated workloads.
Encryption in transit (HTTPS) is on by default for both platforms. At-rest encryption of deployment packages is standard. Scaleway does not currently offer bring-your-own-key for function storage; Lambda supports server-side encryption with customer-managed KMS keys.

Trigger surface is the primary migration challenge. If your Lambda functions are triggered by DynamoDB Streams, Kinesis, or EventBridge rules, there is no drop-in equivalent on Scaleway. You will need to redesign the event pipeline using Scaleway Queues (SQS-compatible) plus a scheduler or an alternative trigger mechanism.
Function code is generally portable. If your Lambda is a straightforward HTTP handler or batch processor without deep AWS SDK calls, moving it to Scaleway is mostly a matter of repackaging and updating the Terraform deployment. Functions that use AWS-specific SDKs (Bedrock, DynamoDB, SES) need their dependencies replaced.
Cold start sensitivity. If your application relies on Lambda provisioned concurrency to guarantee latency, evaluate whether Scaleway's cold start behaviour is acceptable or whether a container-based approach (Scaleway Serverless Containers or Kapsule) gives you more control.
Consider whether serverless is still the right pattern. For workloads that migrated to Lambda to avoid managing servers on AWS, managed containers or a small Kubernetes deployment on Scaleway may offer a better fit with European providers, given the maturity gap between Lambda and Scaleway Serverless Functions.