Google Bundles Antigravity Into Gemini Enterprise: Agentic Coding Goes Corporate

Nine months after launching Antigravity as a free public preview, Google has made its biggest enterprise play yet in agentic software development. In an announcement on the Google Cloud blog, the company said that Google Antigravity is now included in eligible Gemini Enterprise subscriptions — bundling the agentic coding platform, the Android Studio AI tooling, and a new family of IDE extensions into a single corporate license with administrative, security, and spend controls built in from day one.

The move signals a shift in how AI coding tools are sold. Rather than treating agentic IDEs as standalone products with their own billing relationships, Google is positioning Antigravity as a feature of the workspace subscription that companies already buy — the same playbook Microsoft has used to distribute Copilot through Microsoft 365 and GitHub. For Google, the bet is that developer productivity becomes a line item in the IT budget rather than an expense individual engineers sneak onto their corporate cards.

What’s actually shipping

The core of the announcement is availability: administrators with Gemini Enterprise Standard, Plus, and Standard Emerging Market licenses can now enable Antigravity and Android Studio for their users directly from the Gemini Enterprise admin console, with broader support “coming soon.”

Three pieces are new or newly expanded:

Antigravity IDE extensions. Launched alongside this announcement, the extensions bring Antigravity’s coding agents into the editors developers already use — Visual Studio Code, Visual Studio (preview), JetBrains IDEs (preview), and Zed (preview). Previously, agentic features required the dedicated Antigravity desktop app; now the agents come to the developer, not the other way around.

Antigravity 2.0 and the CLI. The desktop app, rebranded Antigravity 2.0, and the Antigravity CLI round out the surfaces, so the same governed agent experience follows engineers from IDE to terminal to desktop.

Enterprise identity plumbing. Across all surfaces, admins can enforce corporate identity standards with native support for Workforce Identity Federation (WIF) and Application Default Credentials (ADC), removing the API-key juggling that typically frustrates enterprise rollouts of developer tools.

The FinOps pitch: no prepaid token left behind

The most technically interesting part of the announcement isn’t the IDE integrations — it’s the billing model. Google explicitly framed the design around feedback from three constituencies: developers who want agent access everywhere, governance teams who want security controls and license management, and finance teams who, in Google’s words, “want pooled usage so that no prepaid token ever goes unused.”

The result is a set of cost-management features that read like a response to the biggest complaint enterprises have about AI tooling — unpredictable spend:

• Granular spend thresholds — monthly project-level budget caps set in the Billing console, with per-user and per-team controls rolling out later this year.
• Pooled quotas — shared token pools that let high-demand teams absorb capacity others don’t use, so purchased quota doesn’t sit idle.
• Overage enablement — admins can allow usage beyond the pool to flow into consumption-based rates, capped monthly, keeping developer workflows unblocked instead of hard-stopped.
• Usage metrics — centralized tracking of token consumption, API calls, and developer activity for optimizing AI investment.

The pooled-quota model in particular is a quiet admission of how AI developer tools actually behave in production: usage is spiky and concentrated among power users, and seat-based pricing wastes enormous amounts of prepaid capacity. Google is betting that flexible pooling is the feature that convinces finance departments to approve a fleet-wide rollout.

Security and governance

Antigravity under Gemini Enterprise inherits Google Cloud’s standard security and compliance protections, but the announcement highlights controls aimed specifically at agentic coding’s unique risks — AI agents that can read repositories, execute code, and in Antigravity’s case, control a browser:

• Configurable security policies covering workspace sandboxing, browser access, and MCP server access, keeping agents inside authorized environments.
• Central audit logging enabled with a single toggle, capturing prompts, agent responses, and metadata for compliance reporting.
• Data privacy guarantees under Google Cloud’s Terms of Service, with all agent activity executing within the customer’s cloud boundary.

That last point is the enterprise-critical one. Consumer AI coding tools train on what you give them; Google is explicitly committing that agent activity in the enterprise tier stays inside the customer’s security perimeter with data ownership intact.

Enterprise customers are already on board

Google lined up an unusually broad set of named customers and partners for the launch, which suggests the rollout has been in production testing for a while. Accenture described deploying Antigravity in Gemini Enterprise to “arm our engineers with Google DeepMind’s premier technology” without choosing “between developer speed and enterprise-grade governance.” AirAsia’s CTO emphasized pushing Gemini models beyond engineering into finance, marketing, legal, and HR workflows. CGI, Cognizant, Datamatics, Deloitte, and Wipro all offered similar production testimonials, with several citing the combination of the 2.0 desktop app, CLI, and new IDE extensions as the unlock for embedding agentic engineering across global delivery centers.

Context: from free IDE to enterprise platform

The trajectory is worth tracing. Antigravity debuted in November 2025 as a free public preview — a VS Code fork with agent-first interaction patterns and browser control, powered by the then-new Gemini 3 models. At I/O 2026 in May, Google expanded it from a single product into what it called “a cohesive ecosystem,” integrated with Agent Platform and Gemini Enterprise. Today’s announcement completes that arc: what began as a giveaway designed to seed developer mindshare is now a SKU, governed by admin consoles, fed by pooled token budgets, and audited for compliance.

That arc is the standard playbook for developer tools — land with a free tier, expand capability, monetize through the enterprise — but the speed here is notable. It took roughly nine months from launch to full enterprise packaging, a pace that reflects how hard Microsoft (Copilot), Anthropic (Claude Code), OpenAI (Codex), and Cursor are pushing into the same corporate budgets. Agentic coding is no longer an experiment for these companies; it’s the front line of the enterprise AI war, and Google just made its bid considerably harder to ignore.

What to watch

Two open questions follow from this launch. First, adoption friction: Antigravity’s agents are powerful but opinionated, and enterprises standardized on Copilot or Claude Code won’t switch on pooled quotas alone. Google’s counter is surface ubiquity — meeting developers in VS Code, JetBrains, and Zed rather than demanding a new editor. Second, pricing pressure: bundling Antigravity into existing Gemini Enterprise subscriptions effectively gives it away to customers already committed to Google’s stack, which may force rivals to respond with their own bundling moves. Either way, the era of AI coding tools as separately licensed side-products looks to be ending — and the enterprise AI subscription just became the battleground.