Overview
Stay on top of your EKS Kubernetes clusters without spending hours handling repetitive tasks. Cast AI automates Kubernetes cost and active optimization in one easy-to-use platform. No more rightsizing recommendations, we replace them by automation.
You will immediately benefit from features like cost monitoring. We will keep your cloud costs in check with smart and powerful Kubernetes automation, including the fastest autoscaling, bin packing, rightsizing, pricing arbitrage, and spot instance management.
Proven with clients around the world, we will bring 50 to 75% average savings. The best thing: it comes with full AI automation so that you don't need to do it.
Highlights
- NEW: Migrate live Kubernetes containers- including those running stateful workloads - with zero downtime. Eliminate resource fragmentation, ensure maximum resource utilization and optimal instance selection, while driving substantial cost savings.
- Get realtime cost monitoring by namespace, workload, or any other tags by application + get active and automated cost optimization.
- We replace recommendations by automation, with the fastest cluster autoscaler that includes real-time rightsizing and pricing arbitrage of AWS instances.
Get personalized pricing in minutes - New
Details
Features and programs
Buyer guide

Financing for AWS Marketplace purchases
Pricing
Free trial
Dimension | Description | Cost/month |
|---|---|---|
Free | Get unlimited Kubernetes monitoring and cost reduction insights. | $0.00 |
Growth | Up to 4 managed clusters. Up to 500 CPU (charged based on usage) | $1,000.00 |
GrowthPro (deprecated) | Deprecated plan | $1,000.00 |
Enterprise | Unlimited managed clusters. Unlimited CPU (charged based on usage) | $5,000.00 |
Growth (deprecated) | Deprecated plan | $1,000.00 |
Cost Monitoring | Analyze your Kubernetes spending with detailed breakdowns across workloads, namespaces, and allocation groups. | $200.00 |
The following dimensions are not included in the contract terms, which will be charged based on your usage.
Dimension | Description | Cost/unit |
|---|---|---|
Additional hourly charge per managed CPU as defined at cast.ai/pricing | Additional hourly charge per managed CPU as defined at cast.ai/pricing | $0.00694444 |
Consumption Fee | Usage-based charges scaled according to platform consumption beyond base allocations. | $0.01 |
Dimensions summary
Top-of-mind questions for buyers
Vendor refund policy
We do not currently offer refunds.
Custom pricing options
How can we make this page better?
Compliance
Compliance information
Legal
Vendor terms and conditions
Content disclaimer
Delivery details
Software as a Service (SaaS)
SaaS delivers cloud-based software applications directly to customers over the internet. You can access these applications through a subscription model. You will pay recurring monthly usage fees through your AWS bill, while AWS handles deployment and infrastructure management, ensuring scalability, reliability, and seamless integration with other AWS services.
Support
Vendor support
Support via dedicated Slack channel. https://castai-community.slack.com/ or support@cast.ai
Service Level Agreement:
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Standard contract
Customer reviews
Automation has reduced capacity toil and has improved cost control for peak traffic workloads
What is our primary use case?
Among the best features CAST AI offers, our main use case is cost optimization for our GKE cluster without adding operational work. We run media streaming for Xstream and IPTV, so our traffic spikes around big releases and in the evening. CAST AI lets us scale up quickly for peaks and shrink back automatically instead of paying for peak capacity all day.
The second benefit is time. The team spends much less time on node pool management and capacity planning, so we can focus on reliability work.
If I could add one thing, I would like more visibility into why CAST AI made a decision, such as why it chose a particular instance type, evicted a pod, or fell back from spot. That would make it easier to trust the automation for more critical workloads.
How has it helped my organization?
The biggest day-to-day change with CAST AI is that we stopped managing node pools by hand on GKE, and we no longer provision manually. Regarding node provisioning, the CAST AI autoscaler watches for pending pods and adds the right instance type within a minute. We used to resize pools manually or pre-provision for peaks. Now we mostly define the node template and constraints, and CAST AI picks the machine. Our job moved from handling capacity tickets to keeping resource requests accurate.
Regarding bin packing, the Evictor keeps packing pods into fewer nodes and removes empty ones. We also run rebalancing to replace the fleet with a cheaper mix. We have enabled rebalancing daily on the majority of our clusters and weekly on some other clusters. In rebalancing, we have excluded stateful workloads. We have also set PDBs and pre-stop webhooks for rebalancing and for the Evictor, so there is no user impact while rebalancing and bin packing.
Regarding spot with fallback, stateless and batch workloads run on spot. When spot capacity is not available, CAST AI falls back to on-demand and later moves back to spot. That made us confident enough to move more workloads to spot. Critical services stay on on-demand.
CAST AI has impacted our organization positively in three main ways. The first is cost: our Kubernetes cost bill dropped by about 10 to 15%, mostly from spot rebalancing and rightsizing. The second is team time: we spend much less time on node pool management, capacity tickets, and late-night scaling issues. The third is reliability during peaks, such as big releases and evening traffic. We no longer need advanced planning or over-provisioning, because the cluster now scales up and back down on its own with the use of the Evictor.
The savings in team time on capacity management are quite large. Before CAST AI, capacity management took about X hours a week across the team, including resizing pools. Now it takes about YR a week, mostly reviewing the savings reports, tuning node templates, and checking evictions and rebalancing results.
What is most valuable?
The most valuable CAST AI features for us are automatic node rightsizing and bin packing, plus spot instance automation with a fallback to on-demand. The cost dashboard also helps us show savings to management.
Among the features that stand out most to me in CAST AI, the first is the autoscaler together with the node template. Pending pods get capacity within a minute, and we control which instance families and zones are allowed.
Controlling the allowed instance families and zones is especially important for our team. We have special discounts on some instance family types. Zones matter for several reasons. Persistent disks on GCP are zonal, so stateful workloads, such as OpenSearch and other stateful services, have to come back in the same zone as their volume. If the autoscaler adds a node in the wrong zone, the pod stays pending. Cross-zone traffic also costs money, so zone control lets us keep services closer together.
In addition, spot availability varies by zone, so allowing several zones gives fallback more room before it has to go to on-demand.
In addition to the features above, the second standout feature is the workload autoscaler. It adjusts pod CPU and memory requests based on real usage. That fixes our problem of teams over-requesting resources, which in turn makes bin packing work better.
What needs improvement?
I would like CAST AI to explain its decisions, showing clearly why it picked an instance type, evicted a pod, or fell back from spot. The second area is better handling for stateful workloads, alerts, and integrations, although I think it is already configured. I think safer rollout is already available as well.
Beyond those, I think that covers everything.
For how long have I used the solution?
We have been using CAST AI since we ran a POC last year on our pre-prod environment, and we have achieved a lot of savings. In GKE, we have integrated our eight to nine pre-prod clusters to trace the costing impact. Since last September, we have been using CAST AI in both pre-prod and prod.
What other advice do I have?
On a scale of 1 to 10, I would rate CAST AI overall a 10.
I would rate CAST AI a 10 because it makes our life easier when handling node pools on GKE during our peaks and scaling down when the peaks go down. The workload autoscaler also helps us save on cost. That is why I give it a 10.
Automation has reduced manual right-sizing and delivers ongoing cloud cost savings
What is our primary use case?
My main use case for CAST AI is auto right-sizing. As a platform or cloud engineer, I serve many application teams in our company, and we manage almost 50 or 60 Kubernetes clusters. Doing the right-sizing manually would be very painful, so we sought automation, which CAST AI offers, and implemented it in certain clusters.
One example of how we use CAST AI day to day is with one of our biggest applications. It is an internal application that serves end users, a reseller application, and it has around 20 or 100 and 1,001, 1,800 VCPU in the cluster. We implemented CAST AI there so that we can reduce the amount of requested CPU and memory from the workload, which reduces the number of worker nodes that we pay for to the hyperscaler. That is one of the use cases that we currently manage using CAST AI.
CAST AI told us that their customers can always reduce the amount of VCPU they have, and when we optimize using CAST AI, it also reduces ours. The impact is that we can reduce our monthly cloud cost, and we have achieved almost 30% compared to before we implemented CAST AI.
How has it helped my organization?
CAST AI has had a positive impact on our organization, mainly on cost, because it is an optimization tool. Because we use the public cloud, we need to maintain the number of virtual machines that we use there. Because we can do right-sizing, the number of virtual machines we use is reduced compared to before we used CAST AI. As a result, we can reduce the amount of money we pay to the hyperscaler or public cloud company. That is the impact on our organization.
Since implementing CAST AI, our cost reduction has been around 20% up to a maximum of 30% compared to before, though I am not sure of the exact number.
What is most valuable?
Among the best features CAST AI offers, I think the workload autoscaler is the standout. Other than the workload autoscaler, CAST AI also has a node autoscaler, which is great as well. It can substitute Karpenter or the cluster autoscaler that other Kubernetes clusters might have. It also manages everything in one platform, so we can track all the optimization in the CAST AI platform.
CAST AI makes our daily work easier and more efficient. Previously, we did the right-sizing manually. We would look at a certain period, and then manually adjust the request CPU and request memory in every deployment that we manage in all of our Kubernetes clusters. We have thousands of services that we need to adjust. After we implemented CAST AI, it does this automatically, so we no longer need to adjust anything manually. All the right-sizing is done by the CAST AI automation agent. That is very efficient and improves our operational excellence. We can also achieve cost optimization.
What needs improvement?
Regarding how CAST AI can be improved, I think the main area is not technical but related to the company's market penetration. We are the first company in Indonesia to use CAST AI, so CAST AI could be more aggressive in penetrating Indonesia or Southeast Asia in general. I believe this pain point does not occur only in my company. It is a common pain point for other companies that use microservices, specifically Kubernetes clusters. I think this is more of a strategic decision than a technical one.
Regarding the product itself, CAST AI could also accommodate other container orchestration platforms, such as AWS ECS. Currently, CAST AI is only compatible with Kubernetes clusters. It would be good if AWS ECS could also be handled by CAST AI, because we also use AWS ECS clusters.
I have no other improvements to mention for CAST AI. I think that is all.
For how long have I used the solution?
We have been using CAST AI for about two years, starting with exploring it around the middle of 2024. We assessed the technology and ran a proof of concept. After that, because we saw that CAST AI could give us a benefit from cost optimization, we committed to a certain amount of licensing by the end of 2024. We committed for one year, and until now we still commit to a certain license of CPU with CAST AI.
What do I think about the stability of the solution?
CAST AI has been pretty stable for us.
What do I think about the scalability of the solution?
Regarding CAST AI's scalability, it has HPA and a node autoscaler natively, and it works perfectly on our platform. It is good for scalability.
How are customer service and support?
Customer support for CAST AI is also great. They have a dedicated technical team for us, and we hold a weekly checkpoint to review the configuration, the optimization, and so on. It is great.
I would rate the customer support from CAST AI an eight on a scale of 1 to 10.
Which solution did I use previously and why did I switch?
We did not previously use a different solution before CAST AI.
What was our ROI?
We have seen a return on investment with CAST AI. I can only share the ROI number we have, which I think is around five or six.
What's my experience with pricing, setup cost, and licensing?
We purchased CAST AI through the AWS Marketplace.
When it comes to CAST AI pricing, setup cost, and licensing, it is still okay because, for now, for every $1 we pay, around $5 comes back to us, I think. Related to cost, I think it is still okay, as long as the benefit is higher than the committed license.
Which other solutions did I evaluate?
Before choosing CAST AI, I think we tried to explore Kubecost, but I am not sure it has automation. We only had the recommendations, and there was still no automation at that time.
What other advice do I have?
Overall, I would rate CAST AI a nine on a scale of 1 to 10.
My rating of 9 out of 10 for CAST AI is because, in terms of the solution, it answers our problem and the impact is real. It directly impacts cost and directly answers our problem. That is why I chose nine.
My advice for others considering CAST AI is that companies that have not yet adopted automation need to try it or do a proof of concept. If you try it yourselves, you can see the real and direct optimization. After that, you can decide whether you need that kind of automation or not.
Which deployment model are you using for this solution?
If public cloud, private cloud, or hybrid cloud, which cloud provider do you use?
Automated Workload Right-Sizing That Cuts Cloud Costs
Optimized cloud workloads and cut costs while highlighting needs for clearer predictive insights
What is our primary use case?
My main use case for CAST AI is to optimize cloud costs and to see if there are any improved performances, especially automating Kubernetes platforms. I don't use it on a daily basis, but this represents my major use case.
I used CAST AI as a prototype while experimenting with it. I used it for three different approaches: workload right-sizing, intelligent node auto-scaling, and spot instance optimization with respect to cloud costs. I analyzed resource consumption to avoid paying for compute resources that were not being used. With respect to Kubernetes automation, I configured auto-scaling policies so that the platform could automatically provision nodes and optimize workload placement, removing any unnecessary capacity that was being used.
Majorly, I was experimenting with CAST AI to optimize Kubernetes infrastructure and to rebalance the clusters being used on the cloud.
How has it helped my organization?
Regarding cost saving, I could optimize certain cloud utilization costs. That was my major use case for CAST AI—understanding how much cloud resources we were using and where I could cut down costs. With respect to time saving, I monitored certain resource utilization, adjusted certain node capacity, and managed the scaling aspect, which was easily tuned with CAST AI's auto-scaling features. Regarding performance issues, CAST AI helped me allocate resources more efficiently to respond to changing workload demands, which improved the infrastructure utilization and consistency between infrastructures during heavy traffic overload or traffic fluctuation.
I don't have an exact calculation on cost savings, but I could reduce approximately 25 to 30% of cloud cost reduction. I could save nearly 2,000 to 3,000 a month.
What is most valuable?
The best features are the automated node auto-scaling, where Kubernetes worker nodes are based on application demand and select cost-efficient instance types. I used it for some of my EC2 instances. I would also mention the spot instance automation, which features certain fallback mechanisms. It supports fallback to on-demand capacity and offers prediction, which reduces compute costs between workloads and tolerates interruptions. CAST AI also includes Kubernetes security, which helps reduce risk alongside infrastructure optimization. Additionally, it supports multi-cloud Kubernetes support, which is valuable since I use different cloud architectures based on different use cases, enabling consistency across different cloud networks.
I used CAST AI primarily to have automated Kubernetes and cloud cost optimization. I made CAST AI an infrastructure layer. I used CAST AI primarily on AWS EKS, which measured improvement through multiple EC2 infrastructure costs and node utilization, and how I could optimize certain costs.
CAST AI helped me scale more dynamic instances while maintaining excess capacity of these compute resources on the cloud. I also tried to experiment with optimization.
Regarding features, the continuous optimization capability is valuable instead of manually reviewing Kubernetes resource utilization. I can adjust the infrastructure every few weeks, and it continuously evaluates workloads. The combination of cost efficiency, automation, and scalability makes CAST AI valuable and more scalable.
What needs improvement?
CAST AI is a useful platform for automating Kubernetes infrastructure and reducing cloud costs. However, I see improvement needed in cost visibility and reporting. There is a dashboard, but it doesn't showcase predictive scaling. A predictive scaling feature could be added where I can see more transparency around automated infrastructure decisions, especially when nodes and workloads are being replaced. There is a predictive cost involved, and that should be readily seen by CAST AI's dashboard so I could make better infrastructure choices and auto-scale in a better way. I would position myself to focus on better cost reporting and smarter predictive scaling as an added feature which could be a selling point for CAST AI.
Additionally, there is the team saving concept where there could be visibility for different teams.
Regarding understanding CAST AI's use case, I attended AWS re:Invent, where they sent out a demo to use CAST AI and other resources. I downloaded the necessary features and undertook the course to understand most of CAST AI's features. Although there was a demo and educational materials, CAST AI's use case needs to be more user-friendly for new users to understand what CAST AI is, what it is truly useful for, and where it exactly stands. The engagement with new users was limited.
I could watch videos and have understanding about Kubernetes and where I could use CAST AI, but what about a person who would not know about Kubernetes or any scaling features? If a new organization wants to know about CAST AI and its use cases, they might face feasibility issues in understanding CAST AI and its applications. If there is a handbook, playbook, or basic guide showing where CAST AI can work well for them, that would help a lot for new users of CAST AI.
I have mentioned most of the improvements I identified.
For how long have I used the solution?
I have used CAST AI as a launch product, which I was introduced to last year as part of the AWS re:Invent program where they reached out to me to use their services. That is how I have been using CAST AI.
What do I think about the stability of the solution?
There are some stability issues with CAST AI, especially during fluctuating workloads, where I saw that it was predicting not precisely or accurately. However, I feel this occurs only on certain instances, but not majorly on similar instances.
What do I think about the scalability of the solution?
As I mentioned earlier, I did face some scalability or stability issues, but CAST AI does handle a lot of auto-scaling needs, especially with Kubernetes, and that is why I would use CAST AI. There were many workloads on Kubernetes which had to be automated and auto-scaled based on the different infrastructures involved in different projects. CAST AI provided great instances for all of these use cases. My experience with changing workloads, scaling up or down of services was positive.
How are customer service and support?
I never faced any issues since I was using CAST AI as a playground or a black box. I didn't face many issues, and I believe CAST AI has a good customer support team that is always ready to troubleshoot and assist with auto-scaling.
Which solution did I use previously and why did I switch?
Before CAST AI, I used some other cloud-native tools. On AWS, there was Cost Explorer. On Kubernetes, there was Cluster Autoscaler. I used Prometheus and Grafana for monitoring CPU, memory, and cost utilization. These were the tools I used before CAST AI. CAST AI can combine all of these tools into a single platform and provide me with most of the answers for my use cases. I got to know about CAST AI because I attended the AWS re:Invent conference last year, when I started using CAST AI.
What was our ROI?
I already mentioned the cost savings of approximately 25 to 30% reduction in cloud infrastructure investment costs, which was a positive factor for me. Regarding fewer employees, I don't think CAST AI would replace employees at this time. That would be ambitious.
What's my experience with pricing, setup cost, and licensing?
The setup cost is positive, especially considering the cost savings and infrastructure automation it provides. The connection is straightforward with the Kubernetes cluster. I didn't face many issues with configuring IAM permissions or the auto-scaling policies. It can take additional effort, especially in production environments where there are workload requirements and other necessary factors which would make the setup process less straightforward. A guide with recommended configurations or easier troubleshooting would be helpful during the setup process.
Which other solutions did I evaluate?
I used different tools as previously mentioned.
What other advice do I have?
I would rate CAST AI a seven.
There are major improvements that can be made. With respect to auto-scaling, most cloud-based services are capable of showing these features. Since we are moving towards agents and multi-agent orchestration, there are certain features most cloud services provide. The novelty and usability of CAST AI could be replaced. However, where they cannot be replaced is the reason I add seven.
Since I was experimenting with CAST AI as a tool for optimizing core features, I didn't share much of my data with CAST AI, or I made it masked. I didn't have much sensitive data to provide.
I would say CAST AI is accurate and reliable because I saw a major reduction in cost savings for cloud optimization and the compute costs I was facing. That was a reliable resource where I could optimize costs. It was accurate in understanding where it could auto-scale certain tasks where there was traffic fluctuation. I would say it is reliable.
I would advise learning more about CAST AI because it requires some pre-learning to use it. It doesn't come as a free tool or service where you would need no understanding about CAST AI and could just incorporate it.
That wouldn't be a good procedure to adopt CAST AI into an organization. Learn what its best use case is, or you would majorly lose money if you don't want to use CAST AI in your use case but are still using it. If you don't have many instances on cloud-based infrastructure or compute instances, or a major Kubernetes auto-scaling need, then using CAST AI for your use case would provide limited value. Think about what use case you are looking into and understand where CAST AI and its features can be incorporated, then make an investment. That is my advice.