A Big Week for AI IDEs, and the Numbers to Match
If you blinked, you missed a lot. In roughly 72 hours this week, Cursor shipped a major new release, opened a bridge to millions of JetBrains developers, and had its revenue figures splashed across Bloomberg. GitHub Copilot rolled out a meaningful enterprise analytics update. And the broader AI coding market continued to post numbers that would have seemed implausible 18 months ago. Here is everything developers need to know right now.
Cursor 2.6: MCP Apps and Team Marketplaces
The headline feature of Cursor 2.6, released March 3, is MCP Apps. MCP Apps support interactive user interfaces like charts from Amplitude, diagrams from Figma, and whiteboards from tldraw directly inside Cursor. Instead of your agent returning a wall of text describing a visualization, it now renders the actual thing inline in your chat panel. Fewer context switches, more flow.
The second addition is Team Marketplaces. On Teams and Enterprise plans, Admins can now create team marketplaces to share private plugins internally, with central governance and access controls. For organizations running standardized toolchains, this is meaningful: instead of each developer manually wiring up the same internal MCP servers, admins push them centrally, the same way you might manage VS Code extensions across a fleet today. The release also ships improvements to core capabilities like Debug mode.
Cursor Lands Inside JetBrains IDEs
The next day, March 4, Cursor announced a move that matters to a large slice of the developer world that has never touched a VS Code-based editor. Cursor is now available in IntelliJ IDEA, PyCharm, WebStorm, and other JetBrains IDEs through the Agent Client Protocol (ACP). Developers who rely on JetBrains IDEs for strong Java and multilanguage support can now use any frontier model with Cursor for agent-driven development.
You need version 2025.3.2 or later of your JetBrains IDE with the AI Assistant plugin enabled. Open the agent selector, select Install from ACP Registry, install Cursor, and start working. You don't need a JetBrains AI subscription to use Cursor as an AI agent. With Cursor ACP, developers can choose frontier models from OpenAI, Anthropic, Google, and Cursor. The Cursor ACP is free for all users on paid plans.
The Agent Client Protocol itself is worth watching as an open standard. Think of it like the Language Server Protocol but for AI agents: implement once, run in any supporting editor. JetBrains and Zed are the two major editors championing it, and the registry of agents is growing quickly.
Cursor Crosses $2 Billion in Annualized Revenue
All of this product momentum is happening against a financial backdrop that underscores how fast this market is moving. Cursor's annualized revenue surpassed $2 billion in February, according to reporting from Bloomberg, reflecting growing adoption of the AI coding assistant. The growth is driven largely by corporate demand rather than individual subscriptions, and revenue has roughly doubled year over year.
The competitive landscape gives that number useful context. As of February, Anthropic's Claude Code has grown to over $2.5 billion in run-rate revenue, and OpenAI's Codex has surpassed more than 1.5 million weekly active users. Microsoft CEO Satya Nadella said in October that GitHub Copilot had more than 26 million users. That is a lot of dollars and eyeballs chasing the same goal: making developers faster. And the agents themselves are getting more capable on that front. Instead of juggling one to three tasks simultaneously, developers can now run 10 or 20 processes concurrently, according to Alexi Robbins, co-head of engineering for asynchronous agents at Cursor.
GitHub Copilot Goes Deeper on Enterprise Metrics
On the GitHub side, the week's news was quieter but operationally significant for teams managing Copilot at scale. On March 2, Copilot usage metrics now includes plan mode telemetry. Enterprises can track adoption and engagement trends for plan mode alongside existing Copilot metrics, enabling a more complete view of how teams are implementing plans directly in their IDE.
Plan mode is worth understanding if you have not used it yet. In plan mode, Copilot analyzes your request, asks clarifying questions to understand scope and requirements, and builds a structured implementation plan before writing any code. The experience is conversational. Having telemetry around it means engineering managers can see whether their teams are actually using the plan-first workflow, rather than guessing. GitHub also made its broader Copilot metrics dashboard generally available on February 27, adding code generation dashboards and organization-level pull request throughput tracking.
What to Do Right Now
Three concrete actions worth taking this week:
- Try MCP Apps in Cursor 2.6. Update to the latest version and look for MCP-compatible tools like Amplitude or Figma. Rendering charts and diagrams directly in your agent chat can cut several context switches out of your day.
- JetBrains users: install the Cursor ACP. If you live in IntelliJ or PyCharm, you now have access to Cursor's agent harness without leaving your editor. It takes a few clicks and costs nothing extra on a paid plan.
- Enterprise leads: check your Copilot metrics. Plan mode telemetry is live. Pull the data and see whether your team is planning before coding. If adoption is low, that is a workflow coaching moment worth acting on.
As tools race to add more concurrent agent processes and richer in-IDE UIs, keep an eye on the cost side of the equation too. Platforms like PorkiCoder let you bring your own API key with zero markup, so you pay exactly what the model provider charges and nothing extra. When the pitch is running 20 concurrent agents, that pricing transparency becomes a real consideration.