TOGETHER WITH STACKLET
The Virtual Token Governance Summit You’ve Been Waiting For (Oct 27th)
AI token usage is growing fast across developer tools, internal agents, and production apps. Most teams can see some of that spend. Few can steer it toward the work that pays off.
Join FinOps leaders, platform teams, and AI builders for 90 minutes on governing AI token usage. Learn where token spend really goes, how to route work to the right model, and how to set budgets that don’t stop developers mid-task.
We’ll close with a first look at Stacklet Token Custodian, the control plane that attributes every AI token and governs it with policy.
Tuesday, Oct 27 | 12:00 to 1:30pm EDT | Free livestream
AI COST OPTIMIZATION
The Token Bill of Materials: A Measured Guide to AI Tokenomics on Google Cloud
Your AI bill can grow even when the price per token falls. An agent may read the same text, call tools, and retry steps many times to finish one task.
This is a useful fix: a “Token Bill of Materials” that shows what each agent spends and why.
In a test using made-up insurance claims, the first design used about 151,000 tokens and cost $0.22 per claim.
Old chat history and a repeated handbook made up 90% of its tokens. The next version used fewer tools, shorter tool results, fresh chats, and shorter answers.
Cost fell 38% to $0.136 per claim, with all 50 test claims still correct. Real claims may be harder, so those results need further testing.
Checks:
Label calls by app, team, or task.
Track costly cases, not just the average.
Set firm spending limits; budget alerts alone do not stop usage.
Include human review, fixes, and upkeep in the business case.
The goal is a lower total cost per correct result, not just fewer tokens.
AI PROVIDERS
Claude Cache Reads Drop to $0.10 per Million Tokens; Asana Cuts Agent Costs 76x
Anthropic
Claude Code cut Sonnet 5.5 cache read prices to $0.10 per million tokens, down from $0.20. Cost tools and budget limits now use the new rate.
Idle compaction in Claude Code shrinks long chats before the one-hour cache runs out, so you waste fewer tokens.
Claude Code now shows cache creation token metrics, and it gives better /usage warnings. It's easier to see what drives spend.
Claude Opus 5.5 and Sonnet 5.5 are now in the Gemini Enterprise app. You pay as you go, so you can compare model costs across regions.
OpenAI
Asana made its browser agent 76x cheaper and 5x faster using Codex. It shows that tuning can cut costs a lot.
Composer predictions in Codex use no credits during beta, so teams can try them for free.
EVENTS
FinOps Weekly Summit 2026
The first sessions from FinOps Weekly Summit 2026 are now available!
Catch the key insights, expert perspectives, and actionable strategies from our speakers.
Watch the sessions now and stay ahead in the evolving world of FinOps.
AI COST MANAGEMENT
One $299 Customer Was Burning $1,840 a Month in AI.
A customer’s AI use can cost more than their whole monthly plan. In this example, a $299 account ran up $1,840 in monthly Amazon Bedrock costs.
The figures are illustrative, but the risk is clear: a shared AWS bill can hide which customers cost more than they pay.
Steps to track and control spending:
Use application inference profiles and cost tags to track spending by product or plan.
Add customer IDs to each AI request and use logs to measure costs by customer and feature.
Build a live spending meter in DynamoDB that reserves funds before each call, then adjusts for the actual cost.
AWS Budgets cannot enforce a live cap for one customer in a shared app. Its billing data arrives too late, and its controls affect shared access. The business fix matters, too.
Set an AI allowance for each plan, warn customers at 80%, and define what happens at the limit.
In the example, fewer repeat calls, caching, and batch work cut monthly AI costs from $1,840 to $610.
Track costs by customer, then use that data to protect profit and set clear limits.
VIDEOS & PODCASTS
AI is making FinOps forget about Cloud
AI is driving cloud costs higher than ever. David Linthicum explains why AI cost management, FinOps, observability, and forecasting are critical to controlling AI and cloud spending while maximizing business value.
AI COST ALLOCATION
How to Validate AI Cost Attribution in Production
Your AI bill may add up, but are the costs linked to the right customers and features?
DoiT’s guide shows how to check those links before using them to set prices or judge profit. Start with one service or feature, not your whole cloud setup.
Then run these checks:
Set the scope, time period, and allowed gap between totals.
Make sure customer or feature IDs stay visible as requests run.
Link those IDs to names that finance and product teams know.
Compare live usage, provider records, and assigned costs for the same period.
Keep costs with no known owner visible.
Test retries, cached results, and shared systems that could change cost per customer.
In the guide’s sample case, not a real customer result, $550 of $10,200 in AI spend had no clear owner.
Fixing a missing customer link and separating batch work brought coverage to 99% for the chosen scope.
Track coverage and gaps over time, with a named owner for each check.
A useful AI cost report needs more than matching totals: it needs proof that each cost belongs where you put it.
AI UNIT ECONOMICS
Total Cost Of AI Is The New Economics Of AI
At the Linux Foundation’s analyst summit, the message was clear: count the full cost, then link it to results.
Tokens make up just 10–25% of total AI spending. The rest includes hardware, power, software, staff, and other costs.
3 actions stand out:
Count all costs. Track the total cost of AI, not just model fees, and work with finance and IT teams to assign spending to the right business units.
Match the model to the task. Summit materials report that 86% of companies are testing or using tools that route tasks to different models.
Make cost data consistent. FOCUS, an open billing data standard, can help teams compare spending across providers and link costs to finance records.
Start with one AI use case.
Track its full cost against a clear result, such as lower support costs or more sales. The goal is not just cheaper AI, but AI that earns its keep.
🎖️ MENTION OF HONOUR
When Token Becomes a Business: Essential AI Investment ROI Calculation Guide for Enterprises
Your AI bill tells you what you spent, not what the work was worth. A token is a small unit of data that an AI model reads or writes.
But counting tokens alone will not show whether AI saves money or helps your team.
Three points stand out:
Track cost per completed task, time saved, and gains in sales, not just token prices.
Plan for agents that make many model calls.
NVIDIA says one agent task may need 50 to 500 calls or more, making costs harder to forecast.
Match the model to the job.
A smaller model may meet the need at a much lower cost, while some tasks need better accuracy or faster replies.
For teams running their own AI systems, include chips, storage, networks, and power in the cost. Finance, tech, and business teams need to share this work.
The goal is not fewer tokens at any cost; it is useful work at a cost the business can justify.
Save 20% on AI Value & FinOps Certifications
The job market is hungry for certified professionals who can prove results. Don't let your company's budget leak because of a lack of specialization.
Use code: FINOPSWEEKLY_20 to get an instant 20% discount on the most prestigious certification bundles:
FinOps Certified Practitioner
FinOps AI Value
FinOps Technology Value
FinOps Certified Engineer
FinOps Certified FOCUS Analyst
AI Economics Community
Join the Slack community where the real AI Money conversations happen
