Optimize Your Cloud Costs and Cut Them by 40%
Get Started For Free

How Activeloop Cut AWS Costs by 50% with Cloudchipr

September 3, 2025
3
min read

About Activeloop

Activeloop builds a Database for AI that simplifies complex data infrastructure and accelerates machine-learning development. The company serves teams that train and iterate on models frequently, with workloads that can swing rapidly based on experiment cycles.

Challenge

Activeloop’s AWS costs were rising due to unmanaged, idle, and orphaned resources - especially sandbox GPU nodes used for training and testing. Engineers often spun up expensive instances for short experiments, but these nodes sometimes remained running after work ended.

Left unaddressed, this behavior would inflate operating costs, reduce budget available for R&D, and create governance gaps across EC2, EBS volumes, Elastic Load Balancers, NAT Gateways, and Elastic IPs. The DevOps team needed clear cost visibility and automated cleanup without slowing down experimentation.

Solution: Cloudchipr Implementation

Activeloop’s DevOps team used Cloudchipr’s no-code automation workflows to identify and stop idle resources across the environment.

What Cloudchipr implemented in AWS

Scope & Onboarding. Activeloop securely connected its AWS accounts via AWS IAM cross‑account roles, and defined idle thresholds, grace periods, working‑hour windows, and exception lists for long‑running AWS resources. After defining the idle thresholds, they created automation workflows in Cloudchipr. These workflows monitored each resource, checking it against its defined idle threshold and subsequently notifying or stopping any resources found to be idle.

Primary AWS services used and how they are used.

  • Amazon EC2 & Auto Scaling - Inventory and state checks to detect compute idle time and shut down/terminate outdated or unused instances via AWS APIs.
  • Amazon EBS - Periodic scans to find unused volumes and snapshots; automated notifications and safe-delete workflows.
  • Elastic Load Balancing (ALB/NLB) - Detection of unattached/low-traffic load balancers for review and removal.
  • Amazon VPC (NAT Gateways, Elastic IPs) - Identification of idle NAT gateways/EIPs to eliminate recurring charges when not needed.
  • Amazon CloudWatch – Source of utilization metrics (CPU, network, GPU %) feeding Cloudchipr analytics for idle/underused resource policies.
  • AWS Identity and Access Management (IAM) – Least-privilege roles enabling Cloudchipr automations to take approved actions.
  • AWS Cost Explorer – Inputs for rightsizing and spend analysis surfaced in Cloudchipr’s dashboards

Operating Model & Guardrails. Automations run on a schedule with notification/approval flows, change windows, and exception lists. Policies are tag‑aware, ensuring the right owner or team receives each alert.

Timeline & Status. Deployment: [May - July 2025]. Production: [August 2025 - Present].

Results & Benefits

Within three months, Cloudchipr’s analytics and workflows delivered a 50% reduction in idle-resource costs with ~$35,000 in quarterly savings. Engineers now receive proactive alerts, and routine cleanup tasks run automatically - freeing time for model development and platform improvements.

Financial stakeholders gained real-time visibility into spend drivers and savings, improving budgeting and accountability without adding process friction.

Customer Quote

“Cloudchipr gave us full visibility into our AWS usage and automated the cleanup of idle resources. It delivered immediate savings and freed up our engineers to focus on building, not managing costs.”

Davit, Co-founder & CEO, Activeloop

Who Cloudchipr helps and how

Cloudchipr is a FinOps platform that unifies cloud cost visualization, analytics, and automation across AWS and other clouds. Teams use Cloudchipr to reveal unused/underutilized resources, automate engineering workflows safely, and align engineering with finance teams on cost outcomes.

AWS designations: AWS Partner (APN)  - AWS Qualified Software.

Conclusion

Cloudchipr helped Activeloop cut idle‑resource spend by 50% (~$35K/quarter) without engineering time spent. Shared dashboards, policy guardrails, and automation workflows established a durable optimization practice on AWS. With these foundations in place, Activeloop can grow their business with AWS without worrying about cost optimization. 

Share this article:
Subscribe to our newsletter to get our latest updates!
Thank you!
Your submission has been received!
Oops! Something went wrong while submitting the form.
Related articles