Has IntelliJ IDEA Lost Its Crown to VS Code?

Preface

For a long time the competition among development tools centered on which IDE was strongest—completion, refactoring, debugging, large‑project indexing, and the ability to handle complex Java/Kotlin codebases. IntelliJ IDEA claimed the Java IDE throne thanks to deep language understanding.

With AI agents entering the development workflow, the relationship between developers and code changes: developers now describe goals, review diffs, verify tests, and adjust direction, while agents write code. Large‑scale semantic understanding, impact analysis, and safe refactoring become more critical.

AI Agent era, editors are no longer the sole entry point

Traditional IDEs bundle editor, language analysis, build system, debugger, refactoring, and code inspection into one product. The Language Server Protocol (LSP) separates the editor (interaction) from the language server (understanding), allowing language capabilities to be reused across tools.

For AI agents, LSP is essential because naive approaches—reading files, full‑text search, feeding large contexts to models—suffer from:

Missing true semantic relationships between variables, classes, and methods.

High token cost from feeding massive file contents.

Confusion over overloaded symbols and cross‑module dependencies.

Unreliable impact analysis during large‑scale refactoring.

LSP provides structured code understanding: an agent can query where a definition lives, who references it, inheritance relationships, rename impact, and current diagnostics without scanning the entire repository.

What JetBrains brings to VS Code

The extension, Java & Kotlin by IntelliJ IDEA, is in preview and exposes IntelliJ’s long‑standing Java/Kotlin analysis to VS Code. It supports:

Java, Kotlin, and mixed projects.

Maven, Gradle, and Bazel builds.

Intelligent completion, navigation, static analysis, and quick‑fixes.

Refactoring capabilities and DAP debugging protocol.

Optimizations for large projects and monorepos.

For VS Code or Cursor users, this combination offers a lightweight UI with familiar extensions while delivering IntelliJ‑level language understanding.

In effect, JetBrains is extracting the most valuable part of IDEA and making it available to other editors and future agent‑driven environments.

It’s more than just autocomplete

Features like Source Action (auto‑organizing imports, quick‑fixes, code generation) and Find All References become crucial for AI agents that need to clean up engineering details after code generation.

A mature programming agent should follow a workflow:

Understand project structure via the language service.

Locate relevant definitions, references, and diagnostics.

Formulate a modification plan.

Execute code changes.

Re‑query the language service and run tests to verify results.

Without LSP, agents rely on raw text similarity; with LSP they obtain a reliable semantic map.

Why JetBrains puts the ability into VS Code

On the surface it expands the ecosystem to VS Code’s large user base. Deeper, it reflects IDE vendors proactively decoupling core language intelligence to serve the AI‑agent era.

Future IDE value will lie in helping humans and agents understand, modify, and maintain massive codebases safely—knowing dependencies, call chains, build boundaries, and refactoring impact.

Opening language intelligence as a service does not mean IDEA is surrendering; it means the core capability is being decoupled from the desktop IDE.

Impact on Claude Code, Codex and similar agents

Terminal‑type agents excel at automation but lack IDE‑level semantic abilities. While Claude Code, Codex, and similar tools can read/write files and run tests, they struggle with large Java/Kotlin projects without structured insight.

LSP fills this gap, allowing agents to query:

All implementations before changing an interface.

Complete references before renaming.

Diagnostics after code changes before running full tests.

Relevant modules in a monorepo.

This improves accuracy and reduces context‑token cost—both critical for enterprise teams.

Subscription and usage threshold

The preview extension is free, but the full Java/Kotlin intelligence will require an IntelliJ IDEA Ultimate subscription after the official release. VS Code or Cursor users will need to budget for this license if they want sustained IntelliJ‑grade capabilities.

For Chinese developers wishing to use Claude Code, Codex, or other agents without overseas payment or network hurdles, the article mentions Code80 as a way to proxy subscriptions to local endpoints.

IDE’s next step: Agent infrastructure

The move signals a redefinition of the IDE concept—from a single window with editor, tree, debugger, etc., to a suite of services: language understanding, refactoring, building, testing, diagnostics, and indexing. Humans can invoke these services in VS Code; AI agents can call them from terminals.

For the Java/Kotlin ecosystem, the real challenge is not generating code but safely changing millions of lines—knowing dependencies, impact, and ensuring refactor safety. JetBrains’ open language intelligence directly addresses this need.

FAQ

Does this extension mean IDEA is no longer important?

No. It shows JetBrains is turning IDEA’s core language intelligence into a service. The full IDE remains valuable, but its underlying code‑understanding will appear in more editors and agent workflows.

What Java/Kotlin capabilities do VS Code users get?

The preview version already supports Java, Kotlin, mixed projects, Maven, Gradle, Bazel, and provides completion, navigation, static analysis, quick‑fixes, refactoring, debugging, and large‑project optimizations.

How does this relate to AI agents?

Agents need deep project semantics; LSP supplies definitions, references, diagnostics, and refactoring impact, reducing reliance on raw text search and improving accuracy while lowering context cost.

Will the full version stay free?

Currently free in preview; the full feature set will require an IntelliJ IDEA Ultimate subscription after release.

How can Chinese users more conveniently use Claude Code, Codex, etc.?

Installing the JetBrains VS Code extension follows normal marketplace steps. For stable access to Claude, GPT, Gemini, and similar models, Chinese developers can use Code80 to proxy subscriptions and APIs.