Factory just raised $150 million. Cursor may be closing a round that values it near $50 billion. Your engineering teams are already running both. That is not a technology decision — it is the beginning of a platform dependency you haven't fully priced.
Two AI coding companies. One weekend edition of the Wall Street Journal. The stories ran on the same pages, inches apart, and the editors probably thought they were covering competition. What they were actually documenting is a procurement trap forming in real time inside enterprise engineering organizations.
Factory, founded in 2023 by Matan Grinberg after he left a theoretical physics doctoral program at the University of California, Berkeley, announced a $150 million Series C led by Khosla Ventures on April 16. The round, which also included Sequoia Capital, Insight Partners, and Blackstone, values the company at $1.5 billion. Those figures are vendor-reported and unaudited. On the same weekend, reports emerged that Cursor, made by Anysphere, is in discussions to raise approximately $2 billion at a valuation approaching $50 billion — a number that would rank it among the most valuable private technology companies in the world.
These are not parallel stories. They are one story told from two altitudes.
The Copilot Is Already Obsolete
Cursor crossed $2 billion in annualized revenue in February 2026, according to Bloomberg reporting. It built that figure faster than any software-as-a-service company in recorded history, going from $1 billion to $2 billion in annualized revenue in roughly three months. Enterprise buyers, not individual developers, now account for approximately 60 percent of that revenue. Nearly half of the Fortune 500 is reportedly in its customer base.
That growth happened while Cursor was still fundamentally an integrated development environment, a very fast, very capable one, but an IDE. Now the company's own leadership acknowledges the product metaphor is shifting. Cursor 3.0, released in April 2026, replaced the classic IDE layout with what the company calls an agent-first interface built around parallel artificial intelligence fleets. Developers no longer interact with a single assistant; they manage multiple agents running simultaneously. The editor is becoming an orchestration surface.
Factory started at that destination. Its agents, called Droids, handle testing, review, documentation, and deployment, not just code generation. Grinberg describes the company's core thesis as "paving the roads" before the autonomous agents can drive, meaning enterprises need documentation, test coverage, continuous integration and continuous delivery pipelines, and internal tool integrations before the agents can work at full capacity. Customers including Nvidia, Adobe, Ernst and Young, Palo Alto Networks, and Morgan Stanley are reportedly using Droids in production. Factory says its customers see 31 times faster feature delivery and 96 percent shorter migration times — figures that are vendor-reported and unaudited, and should be treated accordingly until independent validation exists.
Model Agnosticism Is a Feature Until It Isn't
Factory's stated differentiator is the ability to route tasks between foundation models, switching between Anthropic's Claude, DeepSeek, or whatever model best fits a given workload, without rebuilding integrations each time the artificial intelligence leaderboard reshuffles. Cursor makes the same claim. Both products market model flexibility as a hedge against dependency on any single provider.
The enterprise buyer who treats model agnosticism as a moat is confusing the vendor's routing architecture with their own organizational freedom. The switching cost is not in the model; it is in the workflow grit that accumulates around whichever tool runs daily standup integrations and owns the pull request history.
The deeper issue is that model agnosticism at the application layer does not solve the platform lock-in problem. It defers it. When an engineering organization runs Droids across its legacy migration workloads and Cursor agents across its greenfield development, and both tools have embedded into Slack, Linear, the CI/CD pipeline, and the backlog system, the switching cost has nothing to do with which large language model is underneath. It is the accumulated workflow integration that makes extraction painful.
Anthropic's Claude Code, launched in May 2025 and now reportedly generating more than $500 million in run-rate revenue for Anthropic, competes from a structurally different position. It is the model provider selling the agent layer. OpenAI's Codex operates from the same position. The frontier model companies are not just AI suppliers to Cursor and Factory; they are racing to own the same enterprise relationship those companies are building.
Enterprise Buyers Are Already Running Multiple Agents
Developers working at large enterprises are not choosing between Factory and Cursor. They are using both, often alongside Claude Code, GitHub Copilot, and whatever the platform team approved last quarter. A JetBrains survey from January 2026 found that 74 percent of developers worldwide had adopted specialized AI tools, with GitHub Copilot, Cursor, and Claude Code each holding meaningful share simultaneously among professional engineers.
That multi-tool reality looks like optionality on a team-level spreadsheet. At the organizational level, it is a fragmented procurement posture that makes contract consolidation harder, security review more expensive, and integration governance nearly impossible to enforce.
The enterprise software industry has run this movie before. During the early cloud migration years, organizations signed agreements with multiple infrastructure vendors simultaneously, telling themselves each served a distinct workload. The eventual rationalization, when it came, cost significantly more than a more deliberate initial architecture would have. The AI coding tooling moment feels structurally similar, except the consolidation pressure will arrive faster because these tools are embedding into engineering workflows at a pace no previous software category has matched.
What Factory's Round Actually Signals
Factory reaching a $1.5 billion valuation is less interesting than who put money in and at what stage. Blackstone's participation is a signal worth noting. Asset managers and infrastructure investors entering the venture rounds of enterprise AI software companies generally indicate an expectation of durable, contracted revenue streams, not speculative growth. Sequoia's continued involvement from the seed stage through the Series C shows conviction held across funding cycles, not a late-stage bet on momentum.
Khosla Ventures board member Keith Rabois joining Factory's board adds a governance dimension. Early-stage venture capital backing morphing into active board presence at the Series C stage typically signals that the investor wants influence over the next phase of company direction, not just a financial return. That is worth watching for enterprise customers who will be negotiating multi-year deals as this company scales its sales motion.
Cursor's reported path toward a $50 billion valuation is a different signal entirely. At $2 billion in annualized revenue and a possible 25x revenue multiple, the market is pricing in a trajectory that assumes continued enterprise seat expansion, successful gross margin improvement through proprietary models like Composer and Composer 2, and defense against both the model-provider competitors and the agent-native challengers like Factory. That is a large number of things to go right simultaneously.
Your engineering organization is almost certainly already paying for more than one AI coding platform. Before the next renewal cycle, ask each vendor for an account of exactly which internal systems their agents have write access to, what data those agents have processed and stored, and what the contractual exit process looks like. Model agnosticism sounds like freedom. What it actually describes is which vendor controls the abstraction layer above your most sensitive development infrastructure. If you cannot answer cleanly, the lock-in has already arrived.
