Presented by

Want to appear here? Talk with us

AI spend is exploding and governance is struggling to keep up.

Non-production environments have become hotspots for unchecked AI experimentation, draining budgets with idle resources that linger long after work ends.

With Gartner forecasting $2.52T in AI spend by 2026, the stakes have never been higher.

Join us to learn how FinOps teams can automate governance, enforce smart policies, and eliminate wasted AI spend without slowing innovation.

Organizations are repeating the same mistakes with FinOps that they made with DevOps 12 years ago. And the results will be just as painful.

Back in 2012, companies knew DevOps mattered but had no clue how to make it work. They hired DevOps teams, bought expensive tools, and launched big transformation programs.

Most failed because they treated DevOps as a technical project instead of a culture change.Today, the same pattern is playing out with FinOps.

The Three Mistakes Happening Right Now

The DevOps teams that succeeded made the right way the easy way. They automated testing, deployment, and compliance so doing the correct thing was also the easiest thing.

Successful FinOps works the same way. Cost visibility shows up where engineers already work - in dashboards, pull request reviews, and planning tools.

Cost-efficient patterns become the default - spot instances are pre-configured, lifecycle policies apply automatically, and rightsizing recommendations appear without anyone asking.

Engineers don't need to think about cost as a separate task. The system is designed so that good cost decisions are the same as good performance and reliability decisions.

The playbook is already already wrote for DevOps 12 years ago - we just need to apply it to FinOps.

Read More

Infrastructure is moving faster than FinOps tools built for the pre-AI era can track. AI workloads shift weekly. Kubernetes adds cost layers by the hour.

Finout Extend 2026 puts the seven companies solving this on one stage, and shows where Finout sits at the center of it.

📅​ Date: June 4th
🕗​ Time: 12:00 PM EST / 18:00 CEST
📍 Online

Subscribe to our Events

Brought by CloudZero Cloud Cost Playbook

RDS for Oracle now supports M8i and R8i instances with License Included pricing, bundling software, support, and managed service costs for simpler Oracle TCO with up to 15% better price-performance.

Lambda now supports scheduled scaling of capacity limits, letting you scale functions down to zero during idle periods — a native cost-control option that needs no custom automation.

New R8idn, R8idb, M8idn, and M8idb EC2 instances are now available, delivering up to 43% better per-vCPU performance over prior generations for memory-heavy and general-purpose workloads.

Read All AWS Updates

BigQuery's billing label for Data Transfer Service is changing on August 11, 2026, from DATA_TRANSFER_SERVICE to data_transfer_service — update your cost filters now to avoid gaps in reporting.

BigQuery added AI.COUNT_TOKENS, giving teams a direct way to estimate and track token usage for LLM queries to better plan and control AI spend.

Read All GCP Updates

No major FinOps-relevant updates this week.

Read All Azure Updates

Jill Kay on FinOps Automation and why Manual FinOps doesn't Scale

Starting June 1st, GitHub Copilot switches from a flat-rate model to consumption-based billing using AI Credits.

Microsoft says they can no longer support the old pricing because Copilot has gotten more complex and costs more to run.

What's included in your plan: Copilot Business stays at $19 per user per month and includes $19 worth of AI Credits (1,900 credits). Copilot Enterprise stays at $39 per user per month and includes $39 worth of AI Credits (3,900 credits). For the first three months (June through August), Microsoft is giving you extra credits to ease the transition - $30 for Business and $70 for Enterprise.

What uses AI Credits: Copilot Chat, CLI, cloud agent, Spaces, Spark, and third-party coding agents all consume credits.

The model you choose makes a huge difference in cost.GPT 4.1 costs $2 per million input tokens and $8 per million output tokens. GPT 5.5 costs $5 per million input tokens and $30 per million output tokens. That's a 2.5x to 3.75x price difference depending on which model your team uses.

Important details: Credits pool across your whole organization, which helps if some teams use more than others. Credits expire each month - they don't roll over. Microsoft added a Preview Bill feature in May so you can see what your costs will look like, though one month notice is pretty tight for budget planning.

Read More

The FinOps toolkit just released version 14 with tools that make it easier to ask questions about your cloud costs and find ways to save money.

The biggest addition is a Copilot Studio agent template that lets anyone ask cost questions in plain English.

New ways to find savings: FinOps hubs now pull in recommendations from Azure Advisor and run checks for common waste. The system looks for stopped virtual machines that still cost money, storage disks nobody is using, and other resources sitting idle.You can add your own custom checks too by dropping query files into a folder.

Planning for reservations and savings plans: A new preview dataset tells you which services qualify for commitment discounts.

Other fixes: The update also fixes issues with multi-currency reporting, government cloud deployments, and Power BI report refreshes.

Read More

Your public cloud dashboard won't show you LLM costs fast enough to stop budget overruns. This matters because LLMs charge by the token, not by the hour like traditional cloud services.

The core problem is timing. AWS, Azure, and GCP aggregate your costs into hourly or daily reports. That works fine for EC2 instances or S3 storage.

But LLMs operate differently. Every word in a prompt costs money, and every word the model generates back costs even more.

What you actually need is request-level visibility. You need to see token counts before and after each LLM call.You need to tag costs by user, feature, or prompt template.You need alerts that fire within minutes, not hours.

The solution requires intercepting requests. A gateway or proxy sits between your application and the LLM provider. It counts tokens accurately for each model's specific rules. It reports costs in real time with the granularity you need to make decisions. Some teams see 40-70% cost reductions just by caching semantically similar requests at this layer.

Read More

The job market is hungry for certified professionals who can prove results. Don't let your company's budget leak due to a lack of specialization.

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

If you want to be in the room where the big decisions are made, you need to be at FinOps X. We’ve secured preferred access for our community:

Director - Global Head of Cloud Economics & FinOps at HSBC

Connect with Professional