Operations | Monitoring | ITSM | DevOps | Cloud

If you’re building or running AI-powered features in production, you need a clear understanding of inference costs. Get it right, and you can turn your AI investments into profitable growth. As Larry Advey, Director of Cloud Platform and FinOps at CloudZero and a member of the FinOps Foundation Technical Advisory Council, puts it: “AI investments will only continue to grow.

Earlier this month, we announced the CloudZero Claude Code Plugin and the CloudZero AI Hub — the first step toward putting your cloud cost data directly inside the AI tools your team already uses. The feedback from customers was clear. They said engineers and FinOps teams wanted more tools and more ways to get answers from CloudZero without switching context. Today, we’re delivering more.

Snowflake is a cloud data platform designed for large-scale analytics, data warehousing, and data processing. It allows teams across an organization to run multiple data workloads on a single platform without managing infrastructure. Snowflake’s architecture is also unique. Compute and storage are completely independent and highly elastic. However, Snowflake’s per-second pricing and elastic compute model make costs highly sensitive to usage.

Kubernetes has revolutionized modern computing infrastructure, offering organizations near-infinite scalability, unparalleled agility in deploying new applications, and enhanced security. However, as enterprise cloud adoption continues to accelerate, that agility often comes with an unintended and costly side effect: skyrocketing cloud bills.

For longer than I’m proud of, I was an AI skeptic. Then, over the holidays, I vibe coded an app whose sole purpose was to make me a better person. The app is a motivator. It’s programmed to send me timely reminders along certain themes, like reading every day, making healthy eating choices, and giving myself plenty of time to plan for anniversaries and birthdays.

On March 18, Jensen Huang took the stage at NVIDIA’s GTC conference in San Jose for a keynote that ran well over two hours — covering everything from CUDA’s 20-year history to humanoid robots that may one day wander Disneyland. But buried inside the spectacle was a remarkably clear-eyed articulation of the economic forces now bearing down on every enterprise that builds on cloud infrastructure.

Engineering teams are shipping faster than ever. AI coding tools like Claude Code and OpenAI’s Codex have quietly removed some of the biggest friction points in the development cycle — and the result is that FinOps teams are being asked to keep up with a pace most practitioners haven’t fully reckoned with yet. That acceleration has a cost consequence. More shipping means more services, more experiments, more infrastructure spun up without review cycles.

In 2025, many teams built strong FinOps foundations: These practices created visibility and control. Now it’s time to elevate. FinOps in Action is a three-part series focused on applying that foundation in real engineering scenarios. Each post highlights a different persona and shows how to move from visibility to operational discipline. Today, we focus on Engineering. Engineering teams influence cost through architecture decisions, scaling policies, and workload design.

If your SaaS product handles 1 million chatbot interactions per month, Amazon Lex alone could cost between $4,000 and $7,500. That range assumes current Amazon Lex V2 pricing of about $0.00075 per text request and $0.004 per speech request. Multiply the requests by the rate, and you’re done. Or are you? Conversational AI services rarely behave that neatly in production — and that includes AWS Lex. Amazon Lex is AWS’s conversational AI service for building chatbots.