On August 20, 2026, Google made a structural change to how enterprise teams access AI coding agents: Antigravity, the company’s agentic development tool, is now included in eligible Gemini Enterprise subscriptions. The update arrives alongside new IDE extensions for VS Code, JetBrains, Visual Studio, and Zed, and a set of governance controls designed specifically for engineering organizations that need to manage spending, compliance, and data residency across large developer populations.

The change matters because it shifts the procurement model for AI coding agents from per-developer add-on to bundled enterprise capability, which is the same move Microsoft made with GitHub Copilot and the same move SpaceX-owned Cursor is attempting with its Origin platform. When governance and billing move under one subscription roof, adoption across an organization becomes an administrative decision rather than a grassroots one.

What Changed on August 20

Google announced two simultaneous updates on August 20, 2026, from the Antigravity team and from Google Cloud. They are interconnected.

Subscription bundling. Antigravity is now included in Gemini Enterprise Standard, Gemini Enterprise Plus, and Gemini Enterprise Standard Emerging Market licenses. Administrators enable it from the Gemini Enterprise admin console, where it sits alongside other enterprise AI tools. Developers do not need to manage a separate account, billing instrument, or API key.

IDE extensions. A first set of extensions for VS Code, Visual Studio (preview), JetBrains (preview), and Zed (preview) launched the same day. All extensions support Workforce Identity Federation and Application Default Credentials, so enterprise single sign-on works the same way it does for other Google Cloud services. An employee logs in with their corporate account; the extension detects their license tier automatically.

Spend and quota controls. Administrators can now set monthly project-level budget caps in Google Cloud Billing. Shared token pools let purchased quota flow to the teams that need it rather than sitting idle in a dedicated per-user bucket. Overage billing can be turned on with a spending ceiling so active agent workflows are not cut off mid-task.

Audit and data controls. Audit logging is available with a single toggle and captures prompts, agent tool executions, and generated artifacts. Administrators can set policies for workspace access, MCP server access, and browser access from a single console. VPC Service Controls perimeters are supported for organizations that need to enforce private networking boundaries around AI workloads.

What Antigravity Actually Does

Antigravity is not a code completion tool in the traditional sense. It is built around autonomous agents that can be handed tasks spanning an entire codebase: researching how a function is used, running a local build, browsing the web to check an API specification, or executing a multi-step workflow that touches multiple files, tests, and deploys.

AirAsia cited in the launch announcement that it now generates over 50 percent of its production QA code using Antigravity. That figure describes agents working on real production systems, not sandbox experiments.

The tool ships across several surfaces. The Antigravity 2.0 desktop app is the primary management interface for agents and orchestration. The Antigravity CLI offers a lightweight command-line path. The new IDE extensions let developers work with agents from inside their existing editors without switching to a separate app.

Since the initial Google I/O preview in May 2026, where Google announced Antigravity availability on the Gemini Enterprise Agent Platform, the product has moved from a Google Cloud developer add-on into a capability bundled with the same subscription that includes the Gemini Enterprise app.

Why Enterprise AI Teams Are Watching This

The coding agent market is undergoing a consolidation of procurement models. Until recently, enterprise coding agents required separate subscriptions, separate billing accounts, and separate governance tooling. That friction is collapsing across the industry.

GitHub Copilot was bundled into Microsoft 365 enterprise agreements. Cursor recently launched Origin as an agent-native code hosting layer that competes with GitHub. Claude Code is available inside Azure through the Microsoft Foundry partnership announced in August 2026, as covered in our analysis of Claude on Azure Foundry for enterprise teams. UiPath launched Maestro Flow to bridge the gap between agentic coding prototypes and production deployment, detailed in our UiPath Maestro Flow analysis.

Google’s move bundles Antigravity into an existing enterprise license rather than adding a new procurement line. For a CIO or engineering leader, that changes the approval question from “should we buy this?” to “should we enable this for our developers?” Those are very different conversations.

The spend controls deserve specific attention. Shared token pools are a direct response to a recurring enterprise complaint: organizations purchase AI capacity, it concentrates in a few power users, and other teams treat AI tools as inaccessible because their per-user quota runs out. Pooled quotas flatten that distribution. Monthly project-level budget caps give finance teams a ceiling without requiring per-developer micromanagement.

The Governance Architecture

The following table summarizes the governance controls available in Antigravity within Gemini Enterprise:

ControlWhat It Does
Workforce Identity FederationEnterprise SSO via corporate identity; no separate API key needed
Application Default CredentialsAuthenticated access for agents without manual credential setup
Pooled token quotasPurchased quota shared across team or org rather than locked per user
Monthly budget capsProject-level spending ceiling set in Google Cloud Billing
Overage controlsOpt-in overage with a hard cap so workflows are not cut mid-task
Audit loggingSingle-toggle capture of prompts, tool executions, and outputs
VPC-SC perimeterPrivate networking enforcement for regulated workloads
MCP server policyRestrict which Model Context Protocol servers agents can access
Browser access policyLimit or allow agent browser actuation within defined boundaries
Code privacyNo enterprise code used to train Google base models

The VPC Service Controls integration is particularly important for the sectors Enera works with: financial services, healthcare, and regulated enterprise environments where data cannot leave a defined perimeter, including AI inference traffic.

The Broader Signal

Google is consolidating its enterprise AI developer story around the Gemini Enterprise umbrella. Antigravity, the Gemini Enterprise app, the Gemini Enterprise Agent Platform API, and now the IDE extensions all feed into a single subscription and administrative console.

That unification serves a specific enterprise need: organizations running large engineering teams do not want to manage a matrix of separate AI tool subscriptions, each with its own billing, compliance process, and identity integration. One subscription, one admin console, one audit log is a legitimate competitive proposition against Cursor’s standalone model or Claude Code’s direct API path.

The question for enterprise technology leaders evaluating coding agents is no longer just which model produces the best code. It is which platform integrates with the identity, data residency, compliance, and spend management infrastructure the organization already operates. For organizations running Google Workspace and Google Cloud, the answer just got more straightforward. For organizations on Azure or AWS, the calculus is different, and the vendor landscape now has distinct answers aligned to each cloud.

If you are building or evaluating enterprise AI infrastructure that spans model access, agent governance, and developer tooling, book a call with the Enera team to map the decisions specific to your stack and your compliance requirements.

Sources