Databricks、企業向けAIゲートウェイ「Unity AI Gateway」の一般提供を開始

The last six months have seen a rapid rise in AI-powered productivity and a proliferation of agents across the enterprise. While these agents can transform knowledge work, they also introduce critical challenges at the intersection of cost, control and choice:

• Cost: Unsustainable spend. As adoption scales, AI costs are growing exponentially due to consumption-based pricing models that track tokens rather than seats. To make cost management even harder, AI tools lack robust, native cost controls, and there is no single view of your entire bill.
• Control: Security and IP risks. To be effective, agents must be trusted with your most sensitive data. Yet, models are breaking out of secure sandboxes, supply chain attacks and prompt injections are exfiltrating data, and data retention and memory systems are creating ungoverned data copies.
• Choice: Vendor lock-in. Organizations are under pressure to give developers the freedom to adopt the latest models and tools. The rapid pace of AI innovation (a new model or agent harness every month!) makes it difficult to maintain agility without being tied to specific vendors.

These challenges are further exacerbated by widespread agent sprawl, with organizations attempting to manage multiple AI tools and thousands of custom agents across disjointed platforms.

Enterprise systems have not caught up to these challenges. This is why we are excited to announce the General Availability of Unity AI Gateway. Unity AI Gateway gives enterprises the cost, control, and choice levers they need to manage AI spend and enforce security across AI agents and assets everywhere, whether it is a coding agent, an MCP, an external agent, a skill, or a model. Thousands of customers, like Rivian, Asana, and Edmunds, already use Unity AI Gateway to roll out AI responsibly across their organizations, and over a quadrillion tokens have passed through our gateway in the past year. We at Databricks use Unity AI Gateway internally to manage spend across thousands of employees who each have access to multiple leading AI tools.

Cost: Complete AI observability and cost control

Organizations require complete visibility into where every AI dollar is going, the ability to stop runaway costs in real time, and a way to continuously optimize how their AI budget is used, without slowing down or burdening developers.

Unity AI Gateway serves as the central hub for any AI asset, including external agents, MCPs, skills, and coding assistants, and delivers end-to-end observability and granular cost attribution across models, providers, teams, and applications. By centralizing this data in Unity Catalog, we provide out-of-the-box dashboards and Genie-powered analysis, giving you deep, immediate insights into your AI investment. Furthermore, teams can establish proactive budgets and enforce hard spend caps to ensure AI innovation remains predictable and aligned with business goals.

One technique to help control AI spend is to route across frontier models, both proprietary and open source. Smart Routing, now in Beta, helps organizations maximize the value of their AI spend without limiting developer productivity. Unity AI Gateway dynamically routes each request to the right model based on factors such as quality, cost, performance, availability, and budget, reserving the most capable and expensive models for the tasks that need them while routing other work to more efficient options. The result is better value from every AI dollar without forcing developers to compromise on the tools they use or the quality they need. If you’d like to participate in the Beta, please reach out to your account team.

“Unity AI Gateway has handled our production scale of >100B tokens/month with no availability issues, while giving us visibility into AI spends and user-level controls. Rapid access to the latest frontier and open-source models, combined with easily queryable usage tables, helped us quickly evaluate new models and build custom analytics around AI usage.” —Sriram Ravichandran, Technical Lead - Software Engineering, Zepto

"Unity AI Gateway enables us to centrally govern LLMs at scale across our data products and coding agents. By providing a single governed access layer across model providers, along with comprehensive observability and budget controls, it makes it much easier to safely scale AI across our organization." —Max Lautner, Staff Engineer, Technical Lead, STRABAG SE

Control: Govern data and AI through a unified control plane

We believe that data and AI must be governed together. After all, agents interact with your most sensitive enterprise data, and invoke MCP servers and tools to take actions on behalf of users that create data and systems of record. And the produced agent traces are littered with confidential information, PII data, or other secrets that can be inadvertently leaked. Enterprises need a unified way to control what agents can access and do, while maintaining the identity, audit trail, and policies needed to govern every action.

With Unity Catalog and Unity AI Gateway, organizations can govern data and AI in one place. Unity Catalog provides the identity, permissions, lineage, and auditing that govern access to data and AI assets, while Unity AI Gateway enforces runtime guardrails and contextual policies across AI interactions. Together, they give teams centralized visibility and control to protect sensitive data, enforce policies, and investigate AI activity without having to stitch together separate governance and security systems.

“With Unity AI Gateway, we finally have one source of truth for how AI coding tools and agents are used across the company. Every request runs through a single control plane, with rate limits, PII guardrails, and cost attribution by user, all on infrastructure we already owned. Next, we're working with Databricks to extend that same visibility and control across the rest of our AI stack. Agents are going to keep getting more autonomous. Having one governed control plane means we're watching one door, not a dozen.” —Saravanan Lakshmanan, Tech Lead, Analytics Platforms, Rivian and Volkswagen Group Technologies (RV Tech)

"At OnePay, being AI-forward means giving our teams broad access to AI while honoring our responsibility as stewards of customers' financial data. Databricks' Unity Catalog is the governance foundation we rely on, and Unity AI Gateway extends that same trust layer across our AI tools, agents, and development harnesses. Our teams can now move faster, with consistent guardrails so we don't compromise on quality, efficiency, or observability." —Nameet Pai, Head of Platform & Data Engineering, OnePay

Choice: Open, multi-AI access without lock-in

Developers want the freedom to choose the models, providers, coding agents, and AI tools that work best for their teams, without being locked into a single AI ecosystem. Unity AI Gateway provides an open, multi-AI approach to governance, extending the identity, permissions, policies, and observability of Unity Catalog wherever teams build.

With Unity Catalog, organizations can register and discover Databricks-hosted and external models, coding agents, MCP servers, and skills in a single source of truth. Unity AI Gateway then enforces governance across interactions with those AI assets, extending consistent permissions, policies, and controls wherever they are used. Existing identity providers and security tools can extend into these AI workflows, giving teams the flexibility to adopt new AI technologies without rebuilding governance or replacing trusted enterprise controls.

Unity AI Gateway also provides native access to frontier models, including Anthropic, OpenAI, Gemini, Kimi, GLM, and more, via a single-query API. This makes it incredibly easy to swap models and stay up to date with the latest and greatest.

“Unity AI Gateway extends our Unity Catalog governance to external model endpoints, allowing us to remain coding-agent agnostic while giving engineers the freedom to use the tools that work best for them. At the same time, it provides centralized oversight and cost controls, with the ability to track AI spend from the individual engineer level up through entire departments.” —Bart Consedine, Head of AI Engineering, Magnite

"Unity AI Gateway gives us the flexibility to choose the right model for every coding task without sacrificing governance. We can move quickly as the AI landscape evolves while maintaining consistent permissions, centralized observability, and confidence in how AI is being used across our engineering teams." —Naveen Rao, CEO, Unconventional AI

"Unity AI Gateway gives us a single place to manage and govern AI traffic across our organization. With built-in observability and standardized coding agents that can securely share governed skills, it gives us the foundation to scale AI adoption with greater confidence."—Greg Rokita, VP of Technology, Edmunds

Get started with Unity AI Gateway

Unity AI Gateway is generally available today, giving organizations a unified approach to AI governance that balances cost, control, and choice as AI adoption scales.

To learn more:

• Explore our documentation on AWS, Azure and GCP
• Visit our website
• Register for our AI governance webinar on August 13, hosted by Matei Zaharia, Cofounder and CTO of Databricks, and Kasey Uhlenhuth, Product Director for AI at Databricks