CloudEstimate

Compare Apache Kafka via Confluent Platform across clouds

Extra Small — up to 50 MB/s ingress with short retention windows

Google Cloud

Region us-central1

$4,384.06

$52,609/year

Compute $3,228.06
Storage $1,156.00
Other $0.00

Component summary

  • 3 × n2-standard-16 for Broker
  • 3 × n2-standard-2 for Kraft Controller
  • 2 × n2-standard-2 for Connect
  • 2 × n2-standard-2 for Schema Registry
  • 1 × n2-standard-4 for Control Center

AWS

Region us-east-1

Lowest monthly cost

$2,856.64

$34,280/year

Compute $2,312.64
Storage $544.00
Other $0.00

Component summary

  • 3 × m6i.4xlarge for Broker
  • 3 × m6i.large for Kraft Controller
  • 2 × m6i.large for Connect
  • 2 × m6i.large for Schema Registry
  • 1 × m6i.xlarge for Control Center

Azure

Region eastus

$3,321.36

$39,856/year

Compute $2,505.36
Storage $816.00
Other $0.00

Component summary

  • 3 × Standard_D16s_v5 for Broker
  • 3 × Standard_D2s_v5 for Kraft Controller
  • 2 × Standard_D2s_v5 for Connect
  • 2 × Standard_D2s_v5 for Schema Registry
  • 1 × Standard_D4s_v5 for Control Center

Why the totals differ

AWS is the lowest-cost option for Apache Kafka via Confluent Platform at the extra small tier, landing $1,527.42 per month below Google Cloud. This spread comes from instance and storage pricing in the default regions rather than commitment discounts, so the ranking reflects straight commercial list pricing. Without high availability, the totals stay closer together because there is less replicated capacity, but existing enterprise commitments or hybrid-use discounts could still change the real procurement outcome.

Sources

Reference architecture: Apache Kafka via Confluent Platform, version Confluent Platform current self-managed deployment guidance, retrieved April 13, 2026.

https://docs.confluent.io/platform/current/kafka/deployment.html

Pricing: Google Cloud, AWS, and Azure pricing snapshots retrieved April 12, 2026.

Commercial pricing only. GovCloud, sovereign cloud, and discounts beyond those shown are not modelled.