Presented by

Want to appear here? Talk with us

CloudThrottle helps teams prevent runaway cloud spending with:

Built by cloud architects who’ve delivered real fixed-price projects.

Deep dive article by Jill Kay on FinOps Weekly Blog. Nonprofits that receive AWS credits face a common trap. They treat free credits like free money and build systems without thinking about what happens when the credits run out. Here's a step by step plan by Jill Kay to use AWS credits wisely and avoid that painful transition.

Focus on what you actually need. Don't try every AWS service just because you can. Build only what supports your mission and what you plan to keep running after credits expire.

Build cost controls from day one. The biggest mistake is thinking you can ignore costs because you're not paying yet. That mindset creates expensive problems later. Set up automation now. Schedule dev and test environments to shut down after work hours. Configure S3 to move old data to cheaper storage automatically. Tag every resource with its purpose, owner, and environment.

Use proven blueprints. AWS provides reference architectures built for nonprofits and lean teams. These blueprints follow best practices for cost, scale, and security. Don't reinvent the wheel. Start with what works. You can always adjust later as needs change. Your goal is stability, not complexity.

Plan for life after credits. Don't wait until your last month of credits to figure out what comes next. Build a simple forecast now. Track monthly spending. Identify essential services. Find ways to cut costs by 20 to 30 percent through smarter architecture.

Free credits are a gift, but they're temporary. Use them to build clean systems, automate aggressively, and plan ahead. When credits end, you'll have a lean setup that puts every dollar toward your mission instead of wasted cloud spend.

Read More

Account-Level Tags in Cost Management You can now use Organizations account-level tags directly within Cost Explorer, CUR 2.0, and AWS Budgets, eliminating the need to maintain separate account lists for allocation.

ECS Managed Instances & Spot Savings ECS Managed Instances now support provisioning EC2 Spot capacity, enabling up to 90% savings for fault-tolerant workloads.

Billing Conductor & Telemetry Control New service-scoped custom line items in Billing Conductor allow for precise internal chargebacks. Additionally, CloudWatch can now auto-enable telemetry org-wide for six critical services—though teams should monitor the resulting ingestion costs.

Read All AWS Updates

Granular Pub/Sub Cost Allocation Billing exports to BigQuery now include detailed Pub/Sub usage (snapshot, subscription, topic), allowing for precise per-resource cost analysis.

Vertex AI Pricing Changes Runtime pricing for Vertex AI Agent Engine has been lowered. However, teams should prepare for new metering on Sessions and Memory Bank starting January 28, 2026.

Read All GCP Updates

​FinOps Weekly Call — 2026 Year Launch. We’ll celebrate last year’s wins, set practical goals for the months ahead, and share quick, actionable tactics for saving money and improving visibility across cloud bills.

​Bring your favorite drink and one cost challenge or clever hack to share — there’ll be time for rapid-fire wins and open mic troubleshooting.

​Whether you want to pick up new ideas, show off a success, or help shape the call plans for the year, this session is made for getting things moving (and having a little fun while we do it).

Cloud governance is about to become your biggest competitive advantage in 2026, not because it stops bad things from happening, but because it lets you move faster than everyone else.

Think about Formula 1 racing for a moment. Those cars have the most powerful brakes in motorsport. Not to go slow. They need those brakes so they can hit 230 mph on the straightaways. Without them, drivers would have to crawl around the track to avoid crashing.

Your cloud governance works exactly the same way. The companies moving fastest with AI aren't the ones with no rules. They're the ones with automated systems that make dangerous mistakes impossible, which frees their teams to build without fear.

Gartner says that by 2028, AI agents will make 15 percent of work decisions on their own. That's up from zero percent in 2024. These AI agents won't just write code. They'll spin up servers, optimize resources, and make decisions at machine speed.

There's also a question of who's responsible when AI agents make decisions. You need a clear framework. ML engineers own model training. Developers own permissions and guardrails. Business owners approve solutions after testing. Platform engineers build the governance systems. Executives decide where and how to deploy AI

Your job as a technical leader isn't to know every policy syntax. Your engineers can handle that. Your job is to help other executives understand that governance removes the fear that stops them from saying yes to innovation.

Read More

Welcome to a special episode of the FinOps Weekly Podcast!

This week, we sit down with Alexa Abbruscatto, FinOps thought leader and creator of the FinOps Cashflow Newsletter to explore the evolving world of FinOps, the power of community, and actionable strategies for cloud cost management.

Listen on Apple Podcast - Listen on Spotify - Watch on YT

A cloud engineer got a simple question from their manager: Could switching their Aurora database from I/O Optimized back to Standard save money? The answer turned out to be yes. But getting there meant avoiding a common trap that makes cloud costs look cheaper than they really are.

Aurora Serverless v2 offers two billing options. Standard charges you separately for compute, storage, and I/O requests. I/O Optimized bundles everything together with free I/O but charges more for compute and storage.

In this case, the I/O Optimized option cost 36% more per compute hour and 127% more per GB of storage. The trade-off only makes sense if you're doing enough I/O operations to justify those higher base costs.

Here's why: Aurora scales up and down every second. If your database jumps to 60 compute units for 30 seconds then drops back down, AWS bills you for exactly that spike. But CloudWatch averages that spike across several minutes of data, making your usage look much lower than it actually was.

This is a critical lesson. CloudWatch is built for monitoring performance, not calculating bills. Using averaged metrics to estimate costs will make your spending look artificially low. The solution was to combine two data sources for the same exact time period.

From Cost Explorer: The actual billed amount for compute over 10 days. This is your source of truth because it's what AWS actually charged.

From CloudWatch: The raw usage numbers for storage and I/O operations over those same 10 days. Not averages of costs, but actual counts of operations.

Why 10 days? Cost Explorer only gives you detailed resource-level data for the past 14 days. A 10-day window gives you enough data to see patterns while staying within that limit.

Every single database cluster was overpaying under I/O Optimized. The team saved up to 26% by switching back to Standard.The "optimized" option sounded better but cost more. The standard option turned out to be the better deal once someone actually checked the numbers.

Read More

Great article by Azure Noob. Azure chargeback is not a feature you can turn on in the cloud portal. It is an architecture decision you make before you deploy anything. Most teams get this backwards. They build first, then try to fix it with tags later. That approach fails about 80% of the time. There are only two subscription architectures that reliably support chargeback in Azure.

Model 1: One subscription per application: Each major app gets its own subscription. Sometimes you split it further into dev, test, and production. This works best when teams own their apps from start to finish. It also works well in regulated industries that need strong separation. The downside is you end up with lots of subscriptions to manage. Your reservation discounts also do not pool as well across many subscriptions.

Model 2: One subscription per department Subscriptions align to how your company is organized. Marketing gets one. Engineering gets one. Finance gets one. This matches how your finance team already thinks. They want to know what Marketing spent, not what 47 different apps spent. The catch is you need perfect tag discipline. Every single resource must be tagged correctly. If something is not tagged, that cost cannot be allocated to anyone. You need Azure Policy to force tagging on every new resource. Without enforcement, people forget and your data falls apart.

The Three Step Process

Step 1 is showback. You show departments what they spent using Azure Cost Analysis. No money moves. No one gets charged. You are just building awareness.

Step 2 is cost allocation rules. These only work with Enterprise Agreements or Microsoft Customer Agreements. You virtually redistribute shared platform costs across teams.

Step 3 is true chargeback. You export cost data from Azure and push it into your ERP or finance system. That is where actual internal billing happens. Money moves between department budgets in your financial system, not in Azure.

Azure stops at Step 2. Your finance team owns Step 3.

Most organizations spend 12 to 18 months trying to fix tags in poorly designed environments. Successful ones spend 3 to 6 months restructuring subscriptions first. Then tags fall into place naturally. Pick your subscription model before you build anything, enforce tags from day one, then integrate with your finance systems when you are ready for true chargeback.

Read More

Professional FinOps from the Community

Diana has been super supportive of the FinOps Weekly Initiative, driving people joining the community, even his bf 😄 Knowledgeable pro with community spirit!

Connect with the Professional of the Week

Azure FinOps Cost Optimization Course
(Starting February 2026)

Join The Waitlist

Alfonso is creating a masterpiece. The biggest educational we’ve done till date.

Learn all you need to optimize resources in Azure.

199€ Super Early Bird – Ends December 21st (CLOSED)

⏭ 299€ Early Bird: Opens January 5th – Ends January 11st

499€ Launch Offer: Opens XXX

699€ Starting Offer: Opens XXX

999€ Final Price

The Perfect Training for Azure Practitioners

Participate by answering our questions. There will be surprise gifts to winners every month!