Automated governance is FinOps' next frontier as AI spend spreads beyond engineering
As AI tools put spending power into the hands of sales, finance and executive teams, cloud cost control is no longer just an engineering problem. At FinOps X 2026, Kion senior product manager Tatum Tummins told theCUBE that automated governance is FinOps' next frontier. He argued that organizations need real, policy-driven controls over AI spend, not just dashboards, using soft caps and human-in-the-loop approval workflows to keep innovation from becoming financial liability.
Key Takeaways
- AI is creating a new class of cloud cost challenge that extends beyond engineering teams.
- Kion FinOps+ from Nor Labs Inc. positions itself between enabling experimentation and containing runaway costs.
- Visibility into every dollar is the top priority before optimization and governance.
- Kion's platform is self-hosted and deploys inside a customer's AWS or Azure account for real controls.
- Kion uses soft caps that trigger alerts and human approval workflows rather than hard cutoffs.
- The next step is enforceable controls applied directly through AI providers such as Anthropic, OpenAI, Bedrock and Foundry.
Stats & Key Facts
- #The FinOps Foundation's State of FinOps 2026 Report found 98% of practitioners now manage AI spend.
AI spend spreads beyond engineering
The cost challenge now reaches non-technical teams.
- ›AI tools put spend capabilities into the hands of sales, finance and executive teams.
- ›The State of FinOps 2026 Report found 98% of practitioners now manage AI spend.
- ›Most organizations still lack the guardrails to control that spend at scale.
Tatum Tummins, senior product manager at Kion, framed the tension as one between enabling experimentation and containing runaway costs. He said the first priority is visibility: if he does not have visibility into every dollar in the technology organization, that becomes priority one, and visibility is what enables optimization and governance.
The shadow AI problem
Tummins used a real anecdote to show the stakes.
- ›During an internal hackathon, a sales team member spent more on AI prompts than any developer on the team.
- ›Tools such as Amazon Bedrock and Anthropic now reach non-technical users.
- ›Cost awareness can no longer be assumed across the organization.
Tummins said you have to let folks have a playground but put a fence around it. He warned that teams which think about guardrails and governance now will avoid getting so jaded that they tell everybody to put down their AI tokens, which he said happened earlier with cloud.
How Kion's approach works
Kion offers a self-hosted, policy-driven platform.
- ›The platform deploys directly inside a customer's AWS or Azure account.
- ›It enables real controls over instance types, GPU spend and token thresholds, not just dashboards.
- ›Keeping data inside the customer's authorized environment matters for regulated enterprises.
Rather than hard cutoffs that kill innovation, Kion uses soft caps that trigger alerts and human approval workflows before spend escalates. Tummins described setting a soft cap so that when an engineer hits a token threshold, the team is notified and can decide whether to keep going or pause to discuss.
The next frontier for FinOps
Tummins sees enforceable, provider-level controls as what comes next.
- ›Automated governance is the defining product challenge for the next year.
- ›The next step is moving beyond visibility to enforceable controls at scale.
- ›Controls should be applied directly through AI providers such as Anthropic, OpenAI, Bedrock and Foundry.
Tummins framed this as the AI governance story, focused on making human-in-the-loop approval workflows as easy as possible to implement so organizations can scale comfortably.
Frequently Asked Questions
Why is AI spend a new kind of cost challenge?
AI tools give sales, finance and executive teams the ability to incur spend, so cost control no longer belongs exclusively to engineering, and 98% of practitioners now manage AI spend.
What is the shadow AI problem?
It is the risk that non-technical users run up AI costs unseen, illustrated by a sales team member who spent more on AI prompts during a hackathon than any developer on the team.
How does Kion's platform control spend?
It is self-hosted inside a customer's AWS or Azure account and enables real controls over instance types, GPU spend and token thresholds using soft caps with alerts and human approval workflows.
Why use soft caps instead of hard cutoffs?
Hard cutoffs can kill innovation, so Kion uses soft caps that trigger alerts and human approval workflows, letting teams decide whether to continue spending or pause to review.
What does Tummins see as the next step for FinOps?
Moving beyond visibility to enforceable controls applied directly through AI providers such as Anthropic, OpenAI, Bedrock and Foundry, with human-in-the-loop approval workflows.
Tummins argues that automated governance, built on visibility, soft caps and human approval, is how FinOps teams keep AI experimentation from turning into runaway cost.
Continue Learning
Comments
Sign in to join the conversation