Presented by

Want to appear here? Talk with us

FinOps teams need to understand who is generating that spend, what projects and initiatives it supports, and where that investment is creating value.

AWS has introduced more granular ways to attribute Bedrock costs using IAM principals and session tags. But effective attribution requires more than collecting additional data. It requires building financial context into the way teams access and consume AI services.

Learn how to:

✓ Connect Bedrock spend to the users, teams, projects, and initiatives driving it

✓ Build attribution into existing workflows instead of relying on manual tagging

✓ Turn granular usage data into better reporting, budgeting, forecasting, and chargeback

✓ Create the foundation for evaluating AI spend against the business value it delivers

See what effective AI cost attribution looks like in practice.

Depop's FinOps team built a companion tool using Claude Code to help manage AWS costs, and the results offer a real-world lesson for anyone doing cloud cost work.

The tool runs with read-only access, so it can look at cost data but cannot buy or change anything.

It pulled in cost data, resource details, and team dashboards to build full reports instead of just flagging numbers.

It cut a full day of manual reservation analysis down to minutes, and RDS reservation coverage rose from 23% to over 40%.

It also caught a hidden cost problem: a capacity reservation created during an old incident that kept billing at full price for months after the incident ended.

But the tool was not perfect.

It wrongly flagged a properly running system as wasteful, and it also reported a false cost spike caused by a data pull error.

The team learned that the tool is great at spotting patterns, but a human still needs to check if the finding is real before acting.

Approval steps with finance and team owners stayed just as slow as before, because those checks matter.

AI tools can speed up the thinking part of FinOps work. But the deciding part still needs a person paying close attention.

Read More

Amazon S3 Express One Zone expanded to 7 more Regions, offering up to 10x faster access and 80% lower request costs than S3 Standard.

New burstable T8i EC2 instances launched, delivering up to 30% better price performance than T3 for everyday workloads like CI/CD and small databases.

Amazon ECS Express Mode now supports Graviton (ARM64), giving containerized apps up to 40% better price performance than x86.

Read All AWS Updates

Cloud SQL expanded the C4 machine series across MySQL, PostgreSQL, and SQL Server Enterprise Plus, aiming for better price-performance on demanding workloads.

New troubleshooting guidance for Cloud SQL PostgreSQL helps teams diagnose CPU, memory, and IOPS issues, reducing the risk of overprovisioning.

Read All GCP Updates

Azure is phasing out reservation exchanges for savings-plan-covered services, with exchanges ending February 1, 2027. Existing reservations get just one final exchange, making commitment planning far more important.

Virtual Network Manager now supports mesh connectivity across up to 3,000 VNets, simplifying large network setups without added complexity.

Read All Azure Updates

Join FinOps professionals at the FinOps Weekly Summit 2026 and discover how to:

Transform FinOps from reactive cost control into strategic business value

Build scalable unit economics for the AI era

Learn proven strategies from organizations managing billions in cloud spend

📅 October 20 & 21, 2026 - 9AM ET / 3PM CEST

Discover why organizations need to move beyond cost control and focus on AI business value, cost per task, AI observability, governance, and strategic decision-making.

We speak with Andrée Lundberg, FinOps Lead at H&M Group, about AI Economics, AI cost optimization, FinOps, business value, AI governance, and the future of AI spending.

Kubernetes clusters often pay for pods that do nothing at all. Idle cost is different from inefficient cost.

Rightsizing a pod that nobody uses still leaves you paying for a pod nobody uses. The waste shows up in three places.

Zero-traffic deployments that get shrunk instead of removed, since a smaller idle pod still looks "healthy" on efficiency reports.

Dev and staging namespaces that run all day and night, even though teams only work a fraction of that time.

Orphaned storage and load balancers left behind after a deployment is deleted, quietly billing every month.

Tools like KEDA and kube-downscaler can automate the schedule part safely.

But the savings only count if the cloud provider actually removes the unused nodes.

Stresses guardrails: never auto-delete storage, always scope automation by labels, and check that node hours actually drop, not just pod counts.

For FinOps teams, this is a reminder that dashboards can hide waste as easily as they reveal it.

The biggest savings often come from asking a simple question: is anyone actually using this, right now.

Read More

Cost tags are the backbone of any cloud cost report. But this story shows how a tiny coding mistake can quietly break that trust.

A team added a new CostCenter tag to over 500 cloud roles. The fix looked simple: one line of code.

But those roles already had tags from three different sources, like team defaults, company-wide defaults, and custom overrides.

The team's tagging tool combines these tags in order. Whichever value comes last wins if there is a name clash.

In an early draft, the new CostCenter tag was placed first, so an old tag with a similar name could silently overwrite it.

This is not a big fix that broke. The dashboard would still work.

But cost numbers behind the scenes would be wrong for some roles, and nobody would notice unless they checked by hand.

The team fixed this by testing the tag order in a small sandbox before rolling it out everywhere.

They also kept older, team-specific tags in place so nothing broke.

Small changes to tagging rules can create silent errors in cost reports.

Testing changes on a small scale first, before applying them everywhere, protects the accuracy leaders depend on to make good money decisions.

Read More

Most FinOps programs run into trouble around month seven. The dashboards work, tagging is done, and leaders are happy.

Then a cost spike hits and nobody knows who owns the problem. How can you build a program that lasts beyond the early wins?

A few key ideas stand out. Start with clear roles, not new tools.

A small central team should manage reporting and rate deals, while engineering owns their own fixes and finance owns budgeting.

Without a leader at the top backing the program, the plan usually stalls.

Choose showback or chargeback based on policy, not just data quality. Showback is safer to start with.

Chargeback should be earned once tagging is clean and trust is high.

Build reports for the people who need to act, not just for leadership slides.

Engineers need daily signals, team leads need weekly trends, and executives need monthly summaries tied to business impact.

Automated guardrails work better than manual approval steps. They stop bad spend before it happens instead of cleaning it up later.

Finally, tie every metric to a real person. A number nobody owns will not drive change.

A FinOps program only survives if someone keeps checking it, updating it, and holding people accountable long after the first dashboard goes live.

Read More

Visit FinOps Weekly Hub

Find all Jobs in our FinOps Jobs directory or join FinOps Weekly Jobs Newsletter below to have a weekly email with the most relevant updates every Monday

Join the Slack community where the real FinOps conversations happen

Use code: FINOPSWEEKLY_20 to get an instant 20% discount on the most prestigious certification bundles:

FinOps Certified Practitioner

FinOps Certified Engineer

FinOps Certified FOCUS Analyst

FinOps for AI

Get Certified for Less