How to Apply FinOps Forecasting in Your Company
4 different case studies to learn how to do one of the hardest topics in FinOps
Victor Garcia
October 19, 2025
TOGETHER WITH CLOUDZERO
Use AI? Get Your Costs Under Control Now
Worried about runaway AI costs? We've got you covered.
Our new AI Cost Optimization Playbook is designed to help you:
Identify hidden AI spending before it spirals
Pinpoint cost drivers with actionable insights
Optimize AI usage without slowing innovation
Take control of your AI budget and make every dollar count.
FORECASTING
How to do FinOps Forecasting in your Company
Every company wants to know what their cloud bill will look like next month. But here's the problem: engineering teams work in real time, finance teams think in quarters, and product teams live somewhere in between. Each group asks different questions about the same future.
Finance wants to know if they'll stay within budget. Product teams wonder if they can grow without running out of money. Engineers worry about what happens if traffic suddenly doubles tonight. Even in companies that are good at managing cloud costs, predictions are usually off by 20 to 30 percent.
The article introduces four made-up companies that show how different organizations handle cloud cost prediction.
Then explains several ways to predict cloud costs better. You can look at past trends over 90 days. You can spot weird spikes that don't make sense. You can plan for different scenarios. You can link costs to real business numbers like cost per user or cost per order.
Some creative methods include working backwards from business goals to figure out what infrastructure you'll need. Or using your existing commitments as a starting point. Or watching how your team works, since a 20 percent increase in how fast they deploy code usually means 18 percent more cloud usage two weeks later.
Companies that get better at predicting see real results. After six months, their predictions improve by 25 to 35 percent. They can adjust budgets in two weeks instead of five. They use their committed spending plans 10 to 15 percent better.
The best part happens when prediction stops being one person's job and becomes something the whole company does together. Everyone shares the same dashboards. They predict, measure, and adjust every few weeks.
Finance people learn about how systems work. Engineers learn about money. People feel safe talking about when predictions were wrong.
Good cloud cost prediction isn't about seeing the future perfectly, it's about getting everyone to talk to each other and work from the same information.
CLOUD PROVIDERS
AWS Update Changes Your 2026 Cloud Budget
AWS
Service Lifecycle Changes Impact Migration Plans
AWS announced lifecycle changes across multiple services. Starting November 7, 2025, services moving to Maintenance (including Amazon Cloud Directory, Glacier, and S3 Object Lambda) won't accept new customers. Services like Amazon FinSpace and AWS Proton enter Sunset with ~12-month end-of-support timelines.
Free GenAI FinOps Training on Skill Builder
New one-hour course "Cloud Financial Management: FinOps for Generative AI" now available, plus four refreshed CFM courses. Covers cost management for Amazon Bedrock, SageMaker, EC2 for GenAI, and Amazon Q.
Instant Resource Discovery via Resource Explorer
Immediate resource search now available within Regions without prior setup. Accelerates cost allocation, inventory reconciliation, and audits at no additional charge.
M8a Instances: 30% Better Performance
New EC2 M8a instances deliver up to 30% higher performance and 19% better price-performance versus M7a, with 45% more memory bandwidth across 12 sizes.
and more …
Microsoft Azure
Carbon-Aware API Management
Public preview enables traffic shifting to lower-carbon regions and dynamic load balancing based on real-time emissions. New policy property helps align API workloads with sustainability goals.
Google Cloud
🆕 Oracle OCI
Universal Credits Model
Annual prepaid credits provide predictable OCI spend with discounted pricing. Credits apply across eligible services and regions, with Multicloud support for Oracle Database@AWS/Azure/Google Cloud.
FINOPS EVENTS
Event: The Hybrid FinOps Advantage
FinOps has expanded far beyond public cloud. Managing costs across data centers, AWS, Azure, SaaS, and AI workloads separately prevents total cost visibility and missed savings.
Discover how FinOps 2.0 strategies deliver comprehensive optimization across your entire technology portfolio.
You'll Learn How To:
Achieve total cost visibility across data centers, multi-cloud, SaaS, and AI infrastructure
Optimize the complete technology stack with unified intelligence and automation
Break down silos between FinOps, ITAM, procurement, and engineering teams
Speakers
Jeremy Chaplin, Gerhard Behr & Victor Garcia
November 13th - 6:00 PM CEST / 10AM EST
KUBERNETES
Scaling Kubernetes for AI/ML Workloads with FinOps
Kubernetes has become the go-to platform for running AI and machine learning work in production. However, AI work needs expensive GPU chips that can cost several pounds per hour. Training jobs come in sudden bursts.
Main problems are: GPU waste sits at the top. Storage sneaks up on teams. Auto-scaling sounds smart but often goes wild.
Practical Fixes That Work
Create separate groups of machines for different jobs. Put expensive GPU nodes in their own pool with strict rules about who can use them. Use spot instances for any work that can handle interruptions.
Write automatic cleanup jobs. A simple script can delete storage volumes older than a week or shut down machines that have been empty for ten minutes.
Real Results
One company running 300 training jobs per day on AWS cut their monthly bill from 1.15M$ to under 600k. They did it by tagging every job with a cost owner, moving non-critical work to spot instances, splitting GPUs into smaller slices, and deleting old files automatically. GPU usage jumped from 38 percent to 71 percent while training queue times dropped by half.
📺️ PODCAST
FinOps Best Practices 2025: Culture, Metrics, AI & the Future
We explore how to build a strong FinOps culture, overcome cloud cost optimization challenges, track progress with actionable KPIs, leverage observability, and apply AI to drive smarter cloud financial operations with expert insights from Amit Kinha
SERVERLESS
How to Cost-Optimize AWS Lambda: Step-by-step
A cloud engineer ran a simple test to see if upgrading AWS Lambda functions could save money. The results were clear: switching from Python 3.9 to Python 3.13 and moving from x86 chips to ARM chips cut both speed and cost by about 38%.
The old setup using Python 3.9 on x86 chips took 2,319 milliseconds to run a math-heavy task. The new setup using Python 3.13 on ARM chips took only 1,439 milliseconds for the same work.
For cost, running the function one million times would cost about $4.83 with the old setup and only $3.00 with the new one.
The test was easy to repeat. The engineer created two identical functions with the same memory settings but different Python versions and chip types. Each function ran a simple loop that did one million math calculations. After running each function ten times and checking the logs, the numbers showed consistent savings.
CLOUD PROVIDERS
Beginner’s Guide to Cloud Provider Pricing
Super complete guide on pricing!
Three Main Ways Cloud Companies Charge You
Pay-as-you-go means you only pay for what you use, billed by the second. AWS started this approach, and now all providers offer it.
Reserved pricing gives you big discounts if you promise to use resources for one or three years. AWS and Azure offer up to 72% off, GCP gives 57% off, and OCI provides 30% off.
Spot pricing offers the biggest savings, up to 90% off, by using leftover computer power.
Computer Power Costs
For a basic setup with 2 processors and 8 GB memory, AWS and Azure both charge about $70 per month. GCP costs slightly more at $72. OCI beats everyone at just $39, which is 45% cheaper than AWS.
When you commit for one year, AWS drops to $44 per month, Azure charges $48, and GCP costs $46. For three years, AWS falls to $30 per month, making it the cheapest for long-term commitments.
ARM-based computers offer another way to save. Azure shows the biggest price gap, with ARM costing 65% less than traditional x86 computers.
Storage Costs
For storing 10 terabytes of data, Azure wins at $213 per month, beating AWS's $236. GCP costs $214, while OCI charges $255. As you store more data, Azure keeps its lead with about 9-10% savings over AWS.
🎖️ MENTION OF HONOUR
Slashing Data Transfer Costs in AWS by 99%
AWS charges a lot of money when you move data around. If you send data from one part of their system to another part in the same region but different zones, you pay $0.01 per gigabyte going out and another $0.01 per gigabyte coming in. That means moving one terabyte costs you $20.
The cool part is that uploading to S3 is free, and downloading from S3 within the same region is also free. So instead of sending data directly between two computers in different zones, you can upload it to S3 from the first computer and download it from the second computer. The only cost is storing the data in S3, which is tiny if you delete it right after the transfer.
The author tested this with a full terabyte of data. The normal way would have cost $20. Using the S3 method cost only 8 cents. That's 99.6% cheaper.
There are some limits to know about. S3 files can't be bigger than 5 terabytes. Single uploads can't be bigger than 5 gigabytes without using special multi-part uploads. And this method adds some delay compared to direct transfers.
The reason this works is that Amazon already moves your S3 data between zones behind the scenes. You're just taking advantage of that built-in copying instead of paying for direct zone-to-zone transfers.
PROFESSIONAL SPOTLIGHT
Eric Lam
Leading GCP FinOps
Happy to have Eric Lam supporting the Summit initiative! Great to have one of the leaders in CSP FinOps with us. Thanks for your contributions!
That’s all for this week. See you next Sunday!
2 Days of a Full FinOps Experience
The FinOps Event You Can’t Miss (And it’s FREE)
Move from quick fixes to strategic planning
Strengthen your financial metrics
Use AI to optimize costs
This is the last big FinOps event of 2025, showcasing proven strategies from companies handling large cloud budgets.
October 23 & 24, 2025
3:00 PM - 9:00 PM CEST / 9:00 AM - 3:00 PM EST
Limited seats available
FinOps for Everyone!