AI FinOps exposes the governance gap behind agent sprawl

At a glance

What this is: NHI Mgmt Group's analysis, drawing on AvePoint's blog post: AI FinOps applies FinOps discipline to AI spend, and unmanaged agents, shadow AI, and license overlap are making AI costs hard to explain.

Why it matters: IAM, NHI, and AI governance teams need this because cost visibility without ownership and access context leaves agent sprawl outside effective control.

By the numbers:

• 98% of FinOps practitioners report managing AI costs today, up from just 31% two years ago.
• Gartner projects the average Fortune 500 enterprise will run more than 150,000 AI agents by 2028.

Context

AI FinOps is the discipline of tracking AI spend against business value across tokens, compute, licenses, and agents. It exists because AI costs scale differently from traditional cloud spend, and because finance teams are now being pulled into decisions that used to sit entirely with engineering or IT.

The governance gap is bigger than budgeting. When organisations cannot inventory AI agents, map owners, or see what data those agents can access, they cannot tell whether rising spend reflects value creation or uncontrolled sprawl. That is why AI FinOps is converging with identity and access governance rather than staying a finance-only practice.

Key questions

Q: What breaks when AI agents are not inventoried or classified?

A: When AI agents are not inventoried or classified, organisations lose the ability to assign risk, apply the right obligations, and prove control to auditors. The result is hidden access, inconsistent oversight, and weak accountability when the agent touches regulated data or triggers business actions.

Q: Why do unmanaged AI agents create a larger risk than managed ones?

A: Unmanaged AI agents are harder to audit, revoke, and contain because no one can reliably answer who owns them, what they can reach, or whether they still need access. That makes them more likely to drift, persist after project changes, and become hidden entry points for attackers. Visibility is the first control, because you cannot govern what you cannot see.

Q: How can security teams tell whether AI spend is actually accountable?

A: Spending is accountable when each workload has an owner, a budget, and enforceable limits, and when usage can be attributed without shared secrets or ambiguous service accounts. If the organisation cannot tie consumption to a named application or team, cost governance will remain weak.

Q: Should organisations prioritise accessibility or AI governance first?

A: They should treat them as related governance disciplines with different control surfaces. Accessibility shows whether policy is being sustained in practice, while AI governance shows whether delegated actions and accountability are controlled as automation spreads. The better priority is to fund both as operational capabilities rather than choosing one as a symbolic programme.

Technical breakdown

Why AI spend behaves differently from traditional cloud spend

AI spend is variable because the metering unit is usage, not just provisioned capacity. Tokens, inference calls, and agent-to-agent handoffs can multiply cost even when the underlying architecture looks unchanged. That creates a governance challenge: the same workflow can become more expensive simply because adoption grows, a model changes, or an agent begins chaining other agents into its decision path.

Practical implication: Track AI cost drivers at the workload and agent level, not just at the cloud account level.

How agent sprawl turns cost into an access governance problem

Agent sprawl means the organisation no longer has a small, enumerable set of AI systems. Each agent can carry its own license, compute demand, and access scope, which makes inventory and ownership the real control points. Once agents multiply across Microsoft, Google, and Salesforce environments, spend control and access control become the same governance problem in different language. See the Identity Governance for AI Agents Guide.

Practical implication: Treat every production agent as an identity-bound asset with an owner, a purpose, and a review cycle.

Why shadow AI changes both security exposure and cost accountability

Shadow AI is not only unmanaged spend. It is also unmanaged access, because tools deployed outside formal review can touch data, trigger workflows, and generate incident costs before anyone has mapped them. In that sense, hidden AI spend is often the financial shadow of a wider governance blind spot around non-human access and approval boundaries. See the Shadow AI and AI Agent Discovery Guide.

Practical implication: Build discovery and ownership checks into the same workflow that flags unsanctioned AI usage.

Threat narrative

Attacker objective: The practical objective is not always theft, but uncontrolled access and unaccounted consumption that hides risk and inflates enterprise spend.

1. Entry occurs when employees or teams adopt AI tools or agents outside approved governance, creating shadow AI that finance and security cannot easily see.
2. Credential and access scope then expand as those agents connect to data, workflows, and downstream services without a single owner or policy view.
3. Impact appears as duplicated licensing, inflated token usage, and higher breach costs when unmanaged AI is involved in an incident.

Breaches seen in the wild

• Vercel Context.ai OAuth Supply Chain Breach: Shadow AI app Context.ai OAuth integration exposes Vercel customer data via unmanaged third-party token.
• OmniGPT breach claim 2025: A hacker claims to have leaked 34 million OmniGPT AI chat messages holding users' API keys and credentials; OmniGPT has not confirmed it.

Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

AI FinOps is becoming an identity governance problem, not just a finance discipline. The article shows that AI spend cannot be governed cleanly unless organisations know which agent owns which task, what data it can reach, and who approved it. That moves the control question from budget reconciliation to lifecycle governance. Practitioners should treat AI cost reviews as a proxy for access governance maturity, not as a separate finance exercise.

Shadow AI is now a dual-risk category: unbudgeted consumption and ungoverned access. The same unmanaged tool that creates surprise spend can also create unmanaged data access and incident cost. That is why AI discovery has to happen across both finance and security workflows, not inside a single dashboard. Organisations that split those views will keep seeing cost surprises after the operational mistake has already happened.

Agent sprawl creates a new class of governance debt. Once agents scale into the thousands, ownership gaps, duplicate licensing, and orphaned access decisions become structural rather than exceptional. This is where agentic AI identity becomes the control layer that makes AI FinOps meaningful. Without identity and lifecycle governance for agents, optimisation simply records the sprawl instead of reducing it. Practitioners should manage AI spend as part of the same governance stack that tracks non-human identities.

Cost visibility without access visibility is incomplete. The article makes clear that a cost dashboard can show a spike, but it cannot explain whether an agent should exist, who owns it, or what data it touches. That gap is the core failure mode. Practitioners should align AI FinOps with ownership, approval, and review controls so spend and risk are judged together.

From our research library:

• Organisations with high levels of shadow AI faced an average of $670,000 in additional breach costs compared with those with little or none, according to IBM's 2025 Cost of a Data Breach Report.
• Read next: Identity Governance for AI Agents Guide

What this signals

AI FinOps is converging with identity governance because agent ownership is now a cost-control requirement. The practical signal for programmes is that spend reviews, access reviews, and inventory work can no longer be run as separate cadences. When agents scale quickly, the control that matters is whether each one has a lifecycle owner and an enforceable access boundary, not whether the invoice can be reconciled after month end.

Agent sprawl will become the default budget pressure point in AI programmes. As more teams deploy agents across SaaS and cloud platforms, duplicated capabilities will keep appearing unless governance can compare ownership, purpose, and data access across environments. That means security and finance teams should expect AI cost optimisation to move closer to shadow AI discovery and non-human identity governance rather than remaining a budgeting exercise alone.

For practitioners

• Inventory every AI agent Create a production inventory that covers Copilot Studio, platform-native agents, custom retrieval agents, and any AI workflow built outside IT. Missing agents are the ones most likely to create unexplained spend and invisible access paths.
• Assign a named owner to each agent Make ownership mandatory for approval, billing, and review. If an agent has no named owner, it should not remain in production, because no one will be accountable for its cost or its access scope.
• Review cost and access together Pair finance review with access review so that spikes in token or license spend are evaluated alongside the data an agent can reach and the workflows it can trigger.
• Treat shadow AI discovery as a control function Use discovery to identify unsanctioned agents and tools, then route them into the same governance process used for approved AI systems, including approval, ownership, and retirement. See the Shadow AI and AI Agent Discovery Guide.
• Set budget guardrails before scaling Define spend thresholds, escalation paths, and review triggers before rollout so usage growth does not turn into an unmanaged budget spike that only appears after the fact.

Key takeaways

• AI spend is no longer just a finance problem because unmanaged agents and shadow AI create governance gaps that budgets alone cannot explain.
• The article’s evidence points to rapid scale, with AI cost management now mainstream and agent growth accelerating faster than most governance models.
• The most effective response is to connect cost visibility to ownership, access review, and lifecycle governance for every production agent.

Related NHI Mgmt Group resources: Shadow AI and AI Agent Discovery Guide · Identity Governance for AI Agents Guide · AI Agent Authorisation Guide · Agentic AI Identity Guide

Key terms

• AI FinOps: AI FinOps is the discipline of measuring, attributing, and controlling the cost of AI systems at the level of usage and business value. It extends FinOps into probabilistic workloads where spend depends on model choice, prompt shape, retries, and tool execution rather than fixed infrastructure alone.
• Agent Sprawl: Agent sprawl is the uncontrolled growth of AI agents, scripts, and automation identities across teams and environments. It creates governance strain because each agent can introduce its own permissions, secrets, and ownership gaps, making revocation, review, and accountability harder to sustain.
• Shadow AI: AI agents, copilots, or connected tools operating without full visibility or governance from security teams. Shadow AI becomes an identity problem when those systems authenticate with unmanaged tokens, service accounts, or OAuth apps that can reach production resources.
• AI Governance: AI governance is the set of controls used to discover, classify, approve, restrict, monitor, and revoke AI-enabled access. It connects identity, data, and policy so organisations can manage what AI can reach, what it can share, and when it should be stopped.

Deepen your knowledge

NHI Foundation Level course, the industry's only accredited NHI security programme, covers NHI governance, machine identity security, and identity lifecycle control. It helps practitioners connect identity governance to the broader security and cost-control decisions their programmes now face.