Presented by

Want to appear here? Talk with us

Anthropic kept Opus 4.7's rate card identical to 4.6: $5 input, $25 output per million tokens. But the headline masks a change that will hit your bill directly.

Opus 4.7 ships with a new tokenizer that generates up to 35% more tokens from the same text. Prices did not move. Costs can. On agentic or code-heavy workloads, that gap compounds quickly, and it rarely surfaces until you see an unexpected spike.

We ran the math on three realistic workloads, mapped every discount lever (prompt caching, Batch API), and built a migration checklist so your team can measure the real delta before flipping the switch, not after.

📍 Attending Google Next? Find us at booth #4520. After the show, join Finout & nOps for FinOps After Dark at Prohibition Bar, Las Vegas. Drinks, music, and real talk with the FinOps community. Spots are limited.

RSVP here

A developer running a charity project just learned an expensive lesson about AI model upgrades.

When Google retired the Gemini 2.0 models, the developer switched to Gemini 2.5 Flash Lite, expecting a smooth transition. Instead, costs exploded and response times slowed to a crawl. The model was using massive amounts of processing power to reason through simple requests that didn't need extra analysis. Different models have wildly different minimum "thinking budgets."

The developer had to rewrite code to control these thinking budgets and prevent runaway costs. The fix was switching to Gemini 2.5 Flash with a hard limit of 50 thinking tokens.Response speeds returned to normal and costs dropped back down.

This is a reminder that model upgrades aren't always improvements for your specific use case, and default settings can quietly drain your budget if you're not watching closely.

Code implementation included to prevent from happening again. Great job!

Read More

Brought by CloudZero Cloud Cost Playbook

Amazon Bedrock now attributes model inference costs to IAM principals in CUR 2.0, enabling granular chargeback for AI projects.

EC2 Capacity Manager adds tag-based dimensions to align capacity metrics with cost centers.

OpenSearch Serverless introduced Zstandard (zstd) compression, shrinking index sizes by up to 32% to lower storage spend.

Read All AWS Updates

BigQuery’s new optimized mode for AI functions reduces LLM token consumption and latency for large-scale analytics.

Hyperdisk ML reached GA, offering 2 TiB/s throughput to shorten ML training runtimes and total compute costs.

Read All GCP Updates

Smart Tier for Blob/ADLS is now GA, automatically moving data between tiers based on usage to eliminate manual lifecycle management.

Read All Azure Updates

Most teams don’t set out to build a FinOps tool. It just kind of… happens. The hidden costs of building don't show up in your budget. But in an AI-first world, they're more real than ever.

What we'll dig into:

Join us as we're getting into the real talk behind one of the most common — and quietly costly — calls that engineering and finance leaders make: build or buy your FinOps tooling.

11:00 AM | Erik Peterson - CTO @ CloudZero on Build vs Buy in AI era

11:20 AM | Panel: Erik Peterson, Ben de Mora (FinOps OG) and Victor Garcia

11:40 AM | Q&A with attendees

📅​ Date: May 7th
🕗​ Time: 11:00 AM EST / 17:00 CEST
📍 Online

Me & Ben discuss on building a practice without community support limits results, how inaccurate data erodes engineering trust, and why manual commitment management directly impacts costs.

Listen on Apple Podcast

Listen on Spotify

FinOps is facing an unusual problem: it might succeed itself out of existence. And Frank Contrepois did an amazing analisis on the topic.

Cloud spending is approaching a trillion dollars a year. At that size, cloud stops being just a technology choice and becomes basic infrastructure, like electricity or banking.

When something becomes infrastructure, it attracts rules and standards. This happened with accounting, telecommunications, and power grids. The work doesn't go away, but it spreads out into other departments.

FinOps is starting to follow this same path. Right now, FinOps teams handle cost allocation, cloud commitments, forecasting, and helping engineers make cost-smart choices. They do this work because no other single team owns all of it.

But as these practices become normal, they're moving to other teams. Commitment buying is shifting to procurement. Budget decisions are moving to engineering managers. Cost reports are becoming part of regular finance work. Smart architecture choices are turning into standard engineering practice.

This creates a strange situation. The better FinOps works, the smaller it gets. Not because it failed, but because it succeeded. When something becomes standard practice, it stops being special.

So what's left for FinOps teams to do?

Read More

Build errors are immediate, but cloud cost errors impact budgets weeks later, often leading to wasted spending.

The main issue is timing; costs are only highlighted after they occur. Engineers focus on features, not costs, and decisions are made without immediate financial feedback.

To address this, integrate cost considerations early in the process, like catching typos before sending an email = Shift Left FinOps.

During design, assess potential costs, and in code reviews, include cost estimates. Use consistent tagging to trace costs back to teams.

Implement platform engineering with smart defaults to guide cost-effective choices.

While cost optimization may not be urgent for smaller budgets, establishing good habits early prepares companies for efficient management as spending grows.

Read More

Cloud prices used to go down every year, but now they're going up. AWS, a big cloud provider, cut prices 107 times from 2006 to 2021, but prices have stayed the same or increased since 2016.

For example, storing data in S3 has cost $0.023 per GB since December 2016, even though storage costs have dropped. The cost of computing power went down at first but is now rising again.

AWS says newer services are better value, but many customers don't need the extra speed, so they end up paying more. Prices for block storage haven't changed, and network costs have gone up. AWS will start charging for public IPv4 addresses in 2024. AWS's profit margins grew from 28.5% in 2022 to 37% in 2024, making more money than Amazon's retail side.

Google Cloud is also making a profit. Even though costs are rising, cloud companies are making more money by not lowering prices for customers. In 2024, major cloud providers spent $251 billion on infrastructure, with Amazon planning to spend $200 billion in 2026, mainly for AI. Initially, low prices were supported by other parts of the business, but now that 94% of businesses use the cloud, there's less competition.

Google and Microsoft have raised prices by adding more services. While there are real cost pressures, providers have managed to keep profits high.

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 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:

Our most sad mention in history.

We’ll miss Marcelo positiveness in the ecosystem. A pity we couldn’t met in person. Thoughts with family, friends and Pier Cloud team.

Farewell Post