Amazon Web Services opened public preview access to AWS FinOps Agent on June 9, 2026. The tool takes over a job that used to belong to spreadsheets and tired analysts: reading the AWS bill, spotting what changed, and explaining why. AWS calls it a frontier agent for FinOps practitioners and engineering teams. It answers cost questions in plain English, digs into anomalies without being asked twice, and can open a Jira ticket the moment it finds a rightsizing opportunity worth acting on.
The launch landed at FinOps X 2026, the FinOps Foundation’s annual conference, and AWS backed it up the same week with a post on its Cloud Financial Management blog. The timing tracks with where the money is. Five research firms, including Straits Research, Precedence Research, and Mordor Intelligence, size the global cloud FinOps market between $14.3 billion and $15.6 billion for 2025, climbing to somewhere between $15.8 billion and $17.3 billion in 2026. Cloud waste has already forced companies to staff up dedicated FinOps teams, a trend this publication has covered in depth. AWS is betting an agent can do a chunk of that work automatically, for free, at least for now.
This piece breaks down what the agent does, what it costs during preview, how it stacks up against Cost Explorer, Azure Cost Management, Google Cloud Recommender, and a crowded field of third-party FinOps vendors, and where the agentic cost management race goes from here.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Is AWS FinOps Agent?
AWS FinOps Agent is a managed AI agent built on Amazon Bedrock, and it sits on top of the cost and billing services AWS customers already use rather than replacing them. According to AWS’s own documentation, it is designed to let customers “continuously monitor costs, investigate anomalies, and surface optimization opportunities” across their cloud environments. That is a broader mandate than a dashboard. The agent runs recurring FinOps workflows on a schedule the customer sets, so it can check spend every morning, flag what moved, and hand off a summary before an engineer opens a laptop.
Access starts through the AWS Management Console, and AWS has kept the feature set stable since the June 9 preview launch: natural-language cost questions, automatic anomaly investigation, optimization surfacing pulled from AWS’s existing recommendation engines, and scheduled reporting. It is explicitly labeled preview software, which means AWS reserves the right to change behavior, pricing, and scope before general availability.
How AWS FinOps Agent Investigates Your Cloud Bill
The mechanics matter more than the marketing here. AWS FinOps Agent does not generate cost insight out of nothing. It pulls from services AWS customers already pay for: Cost Explorer for historical spend, Cost Anomaly Detection for flagging unusual charges, Cost Optimization Hub and Compute Optimizer for rightsizing and Savings Plans recommendations, and CloudTrail for correlating a spend spike with the API call or deployment that caused it. Where a human analyst would normally open five browser tabs to connect those dots, the agent does the cross-referencing itself and returns a single explanation.
Once it identifies an actionable recommendation, such as an idle RDS instance or an oversized EC2 fleet, it can open a Jira ticket automatically, tagged to the team that owns the resource. When it detects a cost anomaly, it posts the investigation directly into Slack rather than waiting for someone to check a dashboard. That output-in-your-existing-tools approach is the part AWS is leaning on hardest in its own materials, because the alternative, a new dashboard nobody logs into, is exactly the failure mode FinOps teams have complained about for years.
Sample Prompts and Workflows
Early hands-on writeups, including a walkthrough published on AWS’s community blog, describe interacting with the agent through natural language rather than a query builder. The kinds of questions it is built to answer look like this:
Why did our EC2 spend jump 18% last Tuesday?
Which idle RDS instances can we safely terminate this month?
Generate a cost report for the payments team for the last 30 days.
Open a Jira ticket for any Compute Optimizer recommendation over $500 a month.That list illustrates the interface style AWS and early adopters have described in public writeups. It is not a verbatim transcript from AWS documentation, since AWS has not published an exhaustive prompt catalog.
From FinOps X 2026 to Public Preview: A Timeline
The public rollout followed a tight sequence. AWS unveiled AWS FinOps Agent in a feature preview at FinOps X 2026, the FinOps Foundation’s flagship event, then made the preview generally accessible to AWS customers on June 9, 2026, per its official What’s New announcement. Six days later, on June 15, AWS folded the launch into its AWS Weekly Roundup blog alongside other releases, including Gemma 4 support on Bedrock and Kiro Pro Max.
That is a fast public timeline for a product this consequential. Agent access to billing data, plus the ability to auto-generate tickets, is a meaningfully different trust boundary than a read-only dashboard. AWS has not announced a general availability date. Until that happens, every feature, pricing detail, and integration described in this piece should be read as preview behavior that can still change.
Why AWS Built a Cost Agent Now
Cloud bills have gotten harder to read, not easier. AI workloads spike compute usage in ways that do not follow the predictable patterns FinOps teams built their alerting around, and multi-cloud, Kubernetes-heavy architectures spread spend across services that rarely show up on the same invoice line. This publication has previously reported that cloud waste remains stuck near 29% of budgets even as AI spend pressures finance teams to find savings elsewhere, a gap wide enough that AWS, Microsoft, and Google all have a commercial reason to close it before a rival does.
A secondary summary citing Gartner gives a sense of scale at the top of the funnel: global end-user spending on public cloud services was projected to reach $723.4 billion in 2025, up from $595.7 billion in 2024. That figure covers overall public cloud spend, not the FinOps tooling market specifically, but even a small percentage of waste inside a number that size is a large enough dollar figure to justify AWS shipping a dedicated agent rather than leaving cost control to Cost Explorer alone.
AWS FinOps Agent Pricing: Free Preview, No GA Number Yet
During public preview, AWS is not charging separately for the agent itself. Customers do still pay standard rates for the underlying AWS services the agent touches when it runs a query or generates a report, and AWS has capped usage with a monthly limit it has not fully detailed publicly. AWS has not published what FinOps Agent will cost once it exits preview, and independent coverage of the launch has flagged that silence as one of the bigger open questions for enterprise buyers trying to budget ahead.
That gap matters because most of FinOps Agent’s direct competitors, from CloudHealth to Cloudability to Vantage, typically sell through custom enterprise quotes rather than posted list prices, so a like-for-like price comparison is not really possible yet. What AWS customers can compare today is simpler: free during preview versus whatever a competing platform already charges them.
AWS FinOps Agent vs. Cost Explorer, Azure, and Google Cloud
The closest built-in comparison inside AWS itself is Cost Explorer and Trusted Advisor, both of which show spend and flag best-practice violations but leave the investigating and the ticket-filing to a human. Azure Cost Management and Google Cloud Recommender play a similar role on their respective clouds, useful for visibility, but not built to autonomously chase down why a bill moved and act on it. Third-party players such as CloudHealth by VMware, Vantage, Cloudability (now part of Apptio), Kubecost, and Spot.io each cover pieces of the multi-cloud or Kubernetes-specific cost picture that a single-cloud AWS agent cannot touch. That is likely to remain FinOps Agent’s clearest limitation: it only sees AWS.
| Tool | Vendor | Cloud Scope | Core Function | Autonomous Investigation |
|---|---|---|---|---|
| AWS FinOps Agent | AWS | AWS only | Natural-language Q&A, anomaly root-cause, scheduled workflows | Yes |
| AWS Cost Explorer | AWS | AWS only | Cost visualization and historical reporting | No |
| AWS Trusted Advisor | AWS | AWS only | Best-practice and cost checks | No |
| Azure Cost Management | Microsoft | Azure only | Budgets, cost analysis, recommendations | No |
| Google Cloud Recommender | GCP only | Optimization recommendations | No | |
| CloudHealth by VMware | Broadcom | Multi-cloud | Cost governance, policy automation | Partial |
| Vantage | Vantage | Multi-cloud | Cost visibility, allocation, alerts | Partial |
| Cloudability | Apptio (IBM) | Multi-cloud | Enterprise FinOps, chargeback/showback | Partial |
| Kubecost | Kubecost | Kubernetes | K8s cost allocation and rightsizing | Partial |
| Spot.io | Spot by NetApp | Multi-cloud | Compute optimization automation | Partial |
Readers comparing Kubernetes-specific cost tooling against this list may also want our walkthrough on setting up Kubecost, and teams running workloads on EKS can check our Amazon EKS setup guide for the cluster layer AWS FinOps Agent’s recommendations often touch.
The Cloud FinOps Market by the Numbers
Estimates vary by methodology and forecast window, but the direction is consistent across every firm that covers the space: double-digit growth through the early 2030s as more companies formalize FinOps practices and multi-cloud complexity keeps compounding. None of the five estimates below include the broader $723 billion-plus general public cloud market Gartner tracks. They are specific to FinOps and cost-management tooling itself.
| Research Firm | 2025 Market Size | 2026 Market Size | CAGR | Long-Term Forecast |
|---|---|---|---|---|
| Straits Research | $15.03B | $16.5B | 9.8% (2026-2034) | $34.88B by 2034 |
| Precedence Research | $14.93B | $16.57B | 10.87% (2026-2035) | $41.89B by 2035 |
| MarketsandMarkets | $14.88B | Not stated | 12.6% (2025-2030) | $26.91B by 2030 |
| Mordor Intelligence | $14.39B | $15.77B | 9.56% (2026-2031) | $24.89B by 2031 |
| Fact.MR | $15.6B | $17.3B | 11.0% (2026-2036) | $49.2B by 2036 |
Early Adopters and Real-World Workflows
Since FinOps Agent only went public on June 9, verified customer case studies are still thin on the ground.
What Early Reports Say About Customers
A LinkedIn recap of the FinOps X 2026 announcements, written by conference attendee Omshree Butani, named four organizations reportedly trying the agent early: Workday, AVIV Group, Convera, and Mitre 10. AWS has not published an official case study confirming those deployments, so the list is best treated as an early signal from the conference floor rather than a verified customer roster.
What is confirmed, from AWS’s own materials, is the workflow shape early users should expect. The agent watches spend continuously, and when it flags an anomaly, it can push the finding straight into Slack for the team that owns the resource, cutting out the step where someone has to notice the dashboard changed in the first place.
What Industry Coverage Is Saying
Coverage of the launch has stayed close to AWS’s own framing so far, since the agent is only weeks old. AWS’s Cloud Financial Management team introduced the tool directly: “Today, AWS announces the public preview of AWS FinOps Agent, an agentic AI solution that investigates cost anomalies to root cause and answers cost questions for engineers across your organization, in the tools they already use,” the company wrote in its launch announcement.
“What is AWS FinOps Agent: an agentic AI solution (currently in public preview) that investigates cost anomalies to root cause, answers cost questions across your organization and delivers insights directly in the tools you already use, like Jira and Slack.”
AWS Cloud Financial Management team, “Which AI Tool for Which FinOps Use Case”
On pricing, AWS was direct in a video briefing on the launch: “During this public preview period you can use the AWS agent at no charge.”
Developer-focused coverage has emphasized how little new infrastructure the agent requires. A contributor on AWS’s community blog wrote after testing it: “It is an AI agent that sits on top of your existing AWS cost tools Cost Explorer, Cost Anomaly Detection, Cost Optimization Hub, Compute Optimizer, and CloudTrail and does the manual work you have been doing yourself,” in a post on dev.to.
CIO.com’s coverage of the launch focused on accessibility over capability, noting that “enterprises and developers willing to try out the new FinOps Agent can do so from the AWS Management Console, the hyperscaler said.”
Will Microsoft and Google Answer With Their Own Agents?
As of this reporting, neither Microsoft nor Google has announced a direct competitive answer to AWS FinOps Agent. Azure Cost Management remains a recommendation-and-visibility tool, and Google Cloud Recommender plays the same role inside GCP. Neither has published an agentic version that autonomously investigates anomalies and files tickets the way AWS describes its new agent doing.
That silence probably will not last. AWS, Microsoft, and Google have spent the last two years matching each other’s AI agent announcements almost feature for feature, from foundation-model hosting (Bedrock, Azure AI Foundry, and Vertex AI) to coding agents to now, it appears, cost management. Given how fast that pattern has repeated elsewhere in the AI stack, a comparable FinOps agent from at least one rival hyperscaler inside the next year would fit the established pace rather than break it. That is analysis, not a confirmed roadmap item from either company.
What This Means for FinOps Teams and Cloud Engineers
For FinOps practitioners, the job shifts from hunting for anomalies to supervising an agent that hunts for them first. That is not a small change. Teams that spent years building Cost Explorer dashboards and tagging conventions now need a different skill: judging whether the agent’s root-cause explanation is actually right before a Jira ticket goes to an engineering team that will trust it by default.
Cloud engineers get a more immediate benefit and a more immediate risk in the same package. The benefit is fewer surprise cost postmortems, since Slack alerts on anomalies arrive closer to real time than a monthly billing review. The risk is ticket fatigue if the agent’s rightsizing suggestions are not tuned to a team’s actual traffic patterns, since a wrong recommendation acted on automatically costs more than a wrong recommendation that sat in a dashboard nobody opened.
Risks and Open Questions Around Agentic FinOps
A handful of open questions follow directly from how new this product is. Pricing after preview is unannounced, and companies that build workflows around a free tool risk a cost shock if AWS prices the general availability version around usage or tokens, the way it prices other Bedrock-based agents. Data access is another one. An agent with standing permission to read CloudTrail, billing data, and resource configurations across an account is a meaningfully larger attack surface than a dashboard a person logs into by hand, and AWS has not published detailed guidance yet on scoping that access down.
There is also a single-cloud limitation baked into the product by design. AWS FinOps Agent only sees AWS spend, so any company running workloads across AWS, Azure, and Google Cloud will still need a multi-cloud tool such as Vantage, CloudHealth, or Cloudability to get the full picture, at least until AWS extends the agent’s scope, if it ever does.
Historical Context: From Spreadsheets to Autonomous Agents
FinOps as a formal discipline is barely a decade old. The FinOps Foundation, the nonprofit that runs FinOps X, was established in 2019 to standardize practices that had mostly lived in ad hoc spreadsheets and finance-team tribal knowledge before that. Tooling evolved in stages: first spend dashboards like Cost Explorer, then multi-cloud governance platforms like CloudHealth and Cloudability, then Kubernetes-specific tools like Kubecost as containerized workloads made per-service cost allocation harder to track by hand. Readers who want the mechanics of that layer can see our comparison of Karpenter, Cluster Autoscaler, and KEDA, three of the autoscalers whose decisions directly drive the compute costs FinOps teams chase.
AWS FinOps Agent represents the next stage in that same arc rather than a break from it: an AI system that does not just visualize spend but investigates it and acts on findings inside the tools engineering teams already use. Whether that stage sticks the way dashboards and Kubernetes cost tools did will depend on how the agent performs once real general-availability pricing and broader adoption test it outside the preview window.
Five Predictions for Agentic Cost Management
- AWS will likely announce usage-based or token-based general-availability pricing for FinOps Agent once preview ends, matching how it prices its other Bedrock agents, rather than folding it into existing Support plan tiers.
- At least one of Microsoft or Google ships a comparably autonomous cost-investigation agent within the next 12 months, given how closely the three hyperscalers have tracked each other’s agentic AI releases so far.
- Third-party FinOps vendors lean harder into multi-cloud coverage as their main differentiator, since that is the one gap an AWS-native agent cannot close by definition.
- FinOps job postings start listing “AI agent oversight” or similar language as a required skill within the next year, mirroring how cloud engineering roles absorbed Kubernetes skills a few years earlier.
- Expect at least one public debate over agentic FinOps governance, meaning rules for how much autonomy a cost agent gets before a human has to approve a ticket or a termination, as more companies move past the free preview stage.
Related Coverage
- Cloud Waste Hits 29% as AI Spend Breaks Budgets [2026]
- Step Functions vs Airflow: $0 vs $357/mo Floor [2026]
- How to Set Up Amazon EKS: 12 Steps, 90 Min [2026]
- How to Set Up Kubecost: 12 Steps, 60 Min [2026]
- Google Cloud Hits 82% Growth Ahead of AWS Earnings [2026]
- What Is Amazon RDS?
- Karpenter vs Cluster Autoscaler vs KEDA: 3x Faster [2026]
Frequently Asked Questions
What is AWS FinOps Agent?
AWS FinOps Agent is an AI agent, built on Amazon Bedrock, that AWS put into public preview on June 9, 2026. It answers cloud cost questions in natural language, investigates spending anomalies automatically, surfaces rightsizing and Savings Plans recommendations, and can open Jira tickets or post findings to Slack.
How much does AWS FinOps Agent cost?
It is free to use during the public preview period, subject to a monthly usage limit AWS has not fully detailed. Standard charges still apply for any underlying AWS services the agent queries or acts on. AWS has not announced general availability pricing.
Does AWS FinOps Agent work with Azure or Google Cloud?
No. It only reads and acts on AWS cost and usage data. Multi-cloud shops still need a separate tool such as Vantage, CloudHealth by VMware, or Cloudability to cover spend outside AWS.
How is AWS FinOps Agent different from AWS Cost Explorer?
Cost Explorer is a dashboard for visualizing historical spend. FinOps Agent goes further. It investigates anomalies on its own, answers natural-language questions, and can take action, like opening a Jira ticket, without a person driving each step.
When did AWS launch FinOps Agent?
AWS opened public preview on June 9, 2026, after unveiling it at FinOps X 2026, the FinOps Foundation’s annual conference. AWS covered the launch again in its June 15, 2026 AWS Weekly Roundup blog post.
Which companies are using AWS FinOps Agent?
AWS has not published an official customer list. A LinkedIn recap of the FinOps X 2026 event named Workday, AVIV Group, Convera, and Mitre 10 as early testers, but that comes from a conference attendee’s post rather than an AWS case study.
How big is the cloud FinOps market?
Analyst estimates for 2026 range from about $15.8 billion to $17.3 billion, according to firms including Straits Research, Precedence Research, MarketsandMarkets, Mordor Intelligence, and Fact.MR, with most forecasting compound annual growth above 9.5% through the early 2030s.
Will Microsoft and Google build competing tools?
Neither has announced a direct equivalent as of this writing. Given how closely the three hyperscalers have matched each other’s AI agent launches over the past two years, a comparable tool from at least one competitor within the next year would fit the pattern, though that is analysis rather than a confirmed plan.
