Enterprise finance teams are getting blindsided by AI bills they never approved, can't forecast, and can't explain to their boards. A prepaid token balance budgeted to last the year is gone by Q1. An engineering team's AI coding tool quietly switches to a pricier frontier model mid-billing cycle. An autonomous agent loops overnight, burning thousands of dollars with no human in the loop to notice. This isn't a rare edge case — it's becoming the defining budget crisis of 2026.
1Password just announced a direct answer: AI Spend and Consumption Management, now in Public Preview as part of its SaaS Manager platform. It's the latest signal that enterprise AI is entering its FinOps era, and the organizations that don't build cost visibility now will pay a steep price later.
Here's what enterprise leaders need to understand about the problem, the product, and what it means for your AI strategy.
Why Your AI Budget Is Already Broken (And You May Not Know It)
Traditional SaaS pricing was designed for predictability. You count your seats, multiply by the annual license fee, and the bill is what the bill is. CFOs built their software budgeting processes around this model over two decades. It works cleanly.
AI doesn't work this way. Every API call to Claude, GPT-5.6, or a Cursor-powered coding assistant consumes tokens — and the cost of those tokens varies by model, by input versus output, and by task complexity. There's no fixed per-seat cost you can anchor to. There's no annual renewal moment that forces a budget conversation. Instead, costs accumulate in real time, silently, often across tools that finance never approved in the first place.
The result, as 1Password CFO Greg Henry put it in a VentureBeat interview: "Developers are consuming tokens at a pace that traditional budgets weren't built to manage, and IT and finance teams are being asked to forecast and justify AI investments without a clear view of what's actually driving costs."
What that looks like in practice: a coding agent stuck in a retry loop overnight can consume a week's worth of tokens in a single session. A vendor quietly upgrades your default model to a pricier tier mid-cycle, and every request from that point forward costs more with no change in behavior on your end. A team ramping a new use case doubles your monthly burn rate before anyone thinks to look. And because every AI vendor has its own billing model, its own metrics, and its own dashboard, getting a holistic view means logging into multiple portals, exporting CSVs, reconciling mismatched data, and maintaining a spreadsheet that's already stale by the time Finance asks.
The Goldman Sachs Number That Should Worry Every CFO
Goldman Sachs estimates that token consumption from AI agents alone will grow 24 times by 2030. That's not total AI usage growth — that's specifically the agentic AI traffic growth driven by autonomous systems executing multi-step workflows: booking travel, writing and deploying code, managing customer service interactions, analyzing financial data.
Every one of those workflows generates vastly more API calls than a human typing at a chat interface. An agentic system that completes a five-step process might trigger 50 API calls where a human user would trigger three. At scale, with dozens of agents running across your organization, the token math changes dramatically — and the budget math changes right along with it.
This is why the visibility problem matters now, before the 24x wave arrives. Organizations that wait until agentic AI is fully deployed across their enterprise will find themselves trying to instrument a cost structure that's already out of control.
What 1Password's New Dashboard Actually Does
The AI Spend and Consumption Management capability — currently in Public Preview, with broad availability planned for fall 2026 — is not a standalone product. It's built into 1Password's existing SaaS Manager platform, which already handles application discovery, license management, and spend analytics for over 400 software integrations. Existing SaaS Manager customers get AI Spend and Consumption Management at no additional cost.
At launch, the tool supports three vendors: Cursor, Anthropic (Claude), and OpenAI (ChatGPT and Platform API). Setup is straightforward: connect your vendor admin API keys, and consumption data begins syncing daily into a normalized dashboard. No custom engineering required. No agents to deploy.
What that dashboard gives you:
Unified visibility. A single view of token usage and spend across all supported vendors, eliminating the need to toggle between separate portals with incompatible reporting formats. IT and Finance see the same numbers.
Budget controls with alerts. Organizations can set vendor-level spend limits and configure percentage-based threshold alerts delivered via Slack and email. If your Anthropic prepaid balance drops to 20%, someone gets notified before it hits zero.
Granular attribution. Consumption broken down by team, user, vendor, and model. This is where the real decisions happen — a CFO or CIO can see not just how much was spent, but which team spent it and on which model, and then ask whether that usage is creating business value or needs to be optimized.
Agent-level tracking. The system captures token consumption at the API level regardless of whether a human or an AI agent generated it. Agentic spikes — the ones that can be the hardest to catch before they become expensive — show up in the dashboard alongside human usage.
Critically, this data doesn't just exist in isolation. Because SaaS Manager already has visibility into your broader software portfolio, AI token costs can be contextualized against your total software investment, making the ROI conversation with Finance measurably easier.
The Cloud FinOps Playbook: What's Coming Next for AI Spend
If you've been in enterprise tech long enough to remember the early cloud era, this story sounds familiar.
When AWS, Microsoft Azure, and Google Cloud popularized consumption-based pricing for compute and storage in the 2010s, enterprises initially had no tooling to monitor or optimize their cloud bills. The predictable per-server economics of the data center era didn't translate. Cloud costs were chaotic, fragmented across services, and invisible until the invoice arrived. That gap spawned an entire FinOps ecosystem — companies like CloudHealth, Spot.io, and Apptio built multi-billion-dollar businesses helping organizations understand and govern their cloud spending.
"Consumption-based pricing isn't new," Henry noted. "We saw it arrive with cloud infrastructure, and it took years to build the tools and disciplines to manage it. AI is the next version of that shift."
That analogy has a critical implication: organizations that build AI spend visibility early will have a structural cost advantage over those that scramble to instrument it after they've already scaled. Cloud FinOps maturity became a competitive differentiator — teams that developed cloud cost discipline in 2014 were better positioned than those that built it in 2018. The same dynamic is playing out with AI token economics right now.
For IT and Finance leaders: this is the moment to treat AI spend as a first-class budget category, not a line item buried in developer tooling.
The Shadow AI Problem Is Also a Shadow Cost Problem
Here's the dimension that often gets missed in AI cost conversations: a significant portion of AI tool usage was never sanctioned by IT or Finance in the first place.
According to 1Password's own Access-Trust Gap Report, over 27% of knowledge workers use AI applications their employer didn't approve. Some of that is personal experimentation that never touches corporate billing. But some of it does hit the corporate budget — an engineer spinning up an API key on a shared service account, a team expensing a subscription that finance didn't evaluate, someone provisioning paid workspace seats without going through procurement.
When that happens, IT has no record of the tool, no visibility into what data is entering the AI's context window, and no way to revoke access when the person leaves. Finance has no way to track the spend because it never went through any approval process that would show up in their systems.
This is why 1Password frames the AI spend problem as an identity problem as much as a cost problem. The same platform that governs who has access to what software can now also surface what that software is costing — and flag the tools that exist outside that governance perimeter entirely.
For Technical Leaders: What to Prioritize Now
If you're a CIO, CTO, or Head of AI Engineering evaluating your AI spend posture, here's where to focus:
Inventory your active AI tools first. Before you can control spend, you need to know what's running. That means API keys provisioned across shared service accounts, team-level subscriptions paid through expense reports, and vendor accounts that may have been set up during POC phases and never shut down. Most organizations running this exercise for the first time find two to three times more AI tool usage than they knew about.
Map your token-intensive workflows. Not all AI usage drives the same cost profile. A single agentic workflow running at enterprise scale can generate more token consumption in a day than hundreds of individual chat users in a month. Identify the workflows where token multiplication happens — code generation, document processing, customer service automation — and treat those as your highest-priority instrumentation targets.
Establish attribution before you need to defend spending. The CFO conversation about AI ROI is coming. When it does, the teams that can map token consumption to specific business outcomes — cost per resolved ticket, time saved per engineering sprint, leads processed per hour — will have a fundamentally different conversation than those who can only show a total dollar amount.
Plan for the agentic inflection point. Your current AI token spend may look manageable. Your agentic AI spend in 18 months may not. Build the instrumentation infrastructure now while the numbers are small enough to learn from.
For Business Leaders: The ROI Conversation Your AI Investment Demands
If you're a CFO, COO, or business unit leader who owns budget accountability for AI initiatives, the core challenge is this: AI spending is being approved at the strategic level ("go build with AI") but tracked at the wrong granularity (individual invoices from three vendors, with no connection to business outcomes).
That gap makes it almost impossible to answer the questions boards are increasingly asking:
- Is our AI investment paying off?
- Which teams are getting the most value from the tools we're paying for?
- Where is spending happening that we didn't authorize?
- What does our AI spend look like in 12 months if usage patterns continue?
Visibility tools like what 1Password is building address the first half of this equation — you can see what you're spending and where. But Henry was explicit that spend data alone isn't enough: "To understand whether an AI investment is paying off, that data needs to connect to business outcomes. The challenge is that value looks different across every organization."
That's a systems problem, not just a tooling problem. The organizations that will get AI ROI right are those that connect spend visibility to outcome measurement from the start — defining what "value" means for each use case before authorizing the budget to run it.
For those building their AI governance frameworks now: the spend layer is the necessary foundation. You can't optimize for ROI you can't measure, and you can't measure ROI you can't attribute.
What Happens When Every Agent Has Its Own Budget Line
The agentic AI era introduces a genuinely new governance challenge: the "user" consuming tokens isn't always a person. It's a system. An autonomous agent running on a shared service account generates API calls that look identical to human usage at the API level — but the attribution challenge is fundamentally different.
When the agent is well-governed, tied to a tracked credential within your managed environment, the consumption flows cleanly into your dashboard. When it's running on an unmanaged service account, or using credentials provisioned outside your IT governance perimeter, connecting that consumption back to a specific team becomes difficult — sometimes impossible.
This is where the intersection of identity security and AI spend management becomes genuinely important. The agents running your business in 2027 will need to carry identity just as much as your human employees do. Governing that identity layer now — establishing which agents can use which API keys, at what consumption limits — is the infrastructure work that prevents the attribution problem from becoming intractable at scale.
The Bottom Line
The AI spend crisis is not hypothetical. It's happening right now in finance departments across the enterprise, as token bills arrive that nobody forecasted, from tools IT didn't know existed, for workloads nobody approved.
Goldman Sachs projects 24x growth in AI agent token consumption by 2030. That is not a number you can absorb with a spreadsheet and three vendor portals. It requires the same category of tooling that cloud FinOps required: unified visibility, granular attribution, proactive controls, and governance that spans both the identity and the cost dimensions simultaneously.
1Password's move into this space is an early signal of what's coming — a broader market for AI spend governance that will look, over the next three to five years, very much like the cloud cost management market of the previous decade. The organizations that build the discipline now will spend less, waste less, and be better positioned to justify AI investment to their boards when the conversation gets harder.
The question isn't whether your organization needs AI spend visibility. It's whether you'll build it before or after the next bill arrives.
1Password AI Spend and Consumption Management is currently in Public Preview for existing SaaS Manager customers at no additional cost. Broad availability is planned for fall 2026.
