Set AI budgets and hard spending caps at the user, workspace, or organization level and stop runaway AI spending early
from Kevin Stump
Today we are announcing AI spending controls in the Unity AI Gateway. This release extends Unity AI Gateway’s existing cost transparency proactive budget alerts to give you full control over your company’s AI spend across all your models – from the coding agents your developers use every day, to the production agents serving your customers, to the batch jobs that run overnight:
AI workloads provide disproportionate value – but their cost profile is inherently more difficult to manage than your traditional cloud spend:
- Your nightly call log translation batch job might run fine for a month, but then fail halfway through the work, triggering retry logic that multiplies the cost 10x overnight.
- Your development organization’s coding agents save thousands of developer hours each week—but those same agents make it easy for an engineer to launch an accidental multi-agent experiment on Friday evening that drains the team’s monthly budget by Sunday.
- And with AI leaderboards popping up across organizations, “tokenmaxxing” encourages engineers to consume tokens to get to the top of the charts. What impresses on the best list is less convincing on the bill.
Engineering, support, sales and operations professionals are integrating with AI faster than any other technology in the last decade, unlocking entirely new use cases week after week. However, this adoption presents a management challenge: usage of the base model now spans dozens of teams, hundreds of users, and thousands of agents, with a changing mix of vendors and model tiers. Spend controls need to be consistent across all AI workloads so your business can confidently use AI without worrying about surprises on the bill.
Table of Contents
Configure budget alerts at any granularity
While expense controls need to be consistent, different parts of your business require different cost controls. A platform team takes care of workplace-wide totals. A FinOps leader handles monthly consumption at the organization level. A technical manager takes care of the experimentation budgets per developer. AI Spend Controls lets you set them all from one place and is deeply integrated Databricks’ existing budgets:
- Per user: Set budgets for individual experiments – for example, $2,000 per user per month for technical organization. Catch the developer whose agent is stuck in a loop before it shows up on the profit and loss statement.
- Per use case: Get notified when your organization’s spending on coding agents like Codex or Claude Code exceeds $1,000 per user per month
- Per work area: Hold each unit to its own budget. The production receives $50,000 per month; Sandbox receives $5,000.
- Per account: Set a revenue cap—for example, $200,000 per month for each model, vendor, workspace—and get alerts well before you approach that limit.
And if notice isn’t enough, you can enforce it hard spending caps: Once a budget is exceeded, Unity AI Gateway automatically stops further requests until you increase the limit or the next billing cycle begins.
Get started with Unity AI Gateway budgets today
To track your organization’s AI spending, follow these steps:
Create your Unity AI Gateway budget
- Open your account settings and navigate to use in the sidebar and open it Budgets tab
- Create a budget and select Unity AI Gateway as the resource type
- Optionally, apply the budget to only a subset of the workspaces
- Optionally apply resource tags to configure budgets for a subset of your AI Gateway LLMs. Only AI Gateway LLMs whose tags match your budget tags will count toward the budget. This is useful for configuring use case-specific budgets.
- Configure a “Common Threshold” that sets the monthly spending limit global across all resources in your selected workspaces that match the resource tags
- Configure a “Per User Limit” that sets a monthly spending limit per user in your account
- Configure email addresses to receive notifications when thresholds are exceeded
Once created, pay attention to budget warnings
If one of your budgets is exceeded, you will receive a notification email:
Analyze your active budgets
The Cost You can respond to budget notification emails or proactively monitor the status of your live budgets in the Account section of your account console. On the Budgets On the page you can see at a glance how your budgets are developing:
Open any budget to see how your AI spending is trending:
If you have configured budget thresholds per user level, the Budget Details page will show you how your organization’s users’ individual AI spending is trending. When users exceed their individual threshold, their status and spending are clearly displayed so you can act quickly:
To increase a budget’s threshold, you can simply edit the budget and change its spending limits.
Analyze your company’s AI spending in detail
Unity AI Gateway Budgets give you a comprehensive overview of spending per user and budget. To further analyze which users, models, or use cases are driving your spend, you can leverage Unity AI Gateway’s existing cost tracking capabilities. Each request is logged in the Unity Catalog system tables with a DBU cost, not just the token count. Provisioned throughput, uptime, pay-per-token usage, and even token costs from external model providers are automatically calculated. You can slice the data however your company tracks spending:
- Identity: Aggregate by user or service principal – attribute expenses to the people and systems that generate them.
- Workspace, endpoint and tags: Group by team, environment or cost center.
- Model and provider: See which models (Opus vs. Sonnet) and vendors (Anthropic vs. OpenAI vs. Open Source) drive costs.
- Request tags: Dynamic attribution for SaaS platforms that send proxies to end customers.
Access the Cost Analytics dashboard by navigating to the Unity AI Gateway page in your Databricks workspace and clicking View Dashboard:
This opens a usage and cost analysis dashboard that you can fully customize:
A platform to control data and AI
AI Spend Controls are a natural extension of the governance features you already use in Databricks:
- Unity AI Gateway is your organization’s central AI gateway for managing and accessing LLMs and MCPs.
- Unity catalog is your central catalog for registering and discovering your organization’s data and AI resources. Access permissions, audit trails, and usage data are all available live in Unity Catalog.
- Databricks Budgets form the basis for cost monitoring and warning. With this publication, Databricks Budgets Now you can configure AI-tailored budgets for your organization’s AI workloads.
Databricks gives you a single, consistent system to control what your agents can do, who they can do it for, and how much they can spend on it. Get started today!
https://www.databricks.com/blog/introducing-ai-spend-controls-unity-ai-gateway
