Vibe coding now rivals OpenAI in Latam AI spending

In the AI Spend Monitor's latest snapshot, 15 out of every 100 dollars spent on artificial intelligence by the companies tracked by Clara went to Cursor, Lovable, and Replit. Across the industries in the base, the three platforms combined already move nearly as much as OpenAI's share of 16.9%.

The comparison doesn't turn vibe coding into a single vendor. It reveals the financial size the category has reached. Tools specialized in building software with AI have stopped being a peripheral bet for developers and are now competing for a relevant share of corporate budgets.

Cursor leads this movement. On its own, it accounts for 10.5% of all AI spending recorded in the period. Lovable accounts for 3.6% and Replit for 0.9%. Together, the three platforms form the main block of specialized applications outside the large language models.

More companies pay for Lovable than for Cursor. Even so, total spending on Cursor is almost three times higher. In simple terms, Lovable reaches more companies, but Cursor moves more money among the companies it serves.

One possible explanation lies in how these tools enter the workflow. Lovable lets people create sites and applications through everyday-language commands, which lowers the barrier to getting started. Cursor works as a programming environment used throughout development. If more members of a team start using it, or if it becomes part of the routine on several projects, the company's bill tends to grow. The data doesn't show how many licenses or projects exist, but the gap between adoption and spend is consistent with that kind of more intensive use.

The evolution of Codex and Claude could reduce the need for separate tools, since those products also help write code. But Cursor and Lovable offer more than generating a single response. They organize the creation process within a specific product. For a company, there can be a difference between asking a model to write a snippet of code and using a platform built to turn an idea into software. It's that space between generating code and building a product that vibe coding platforms are trying to occupy.

The data shows that this proposition finds buyers. Cursor, Lovable, and Replit still reach a limited share of companies, but their share of spending is far larger than their presence would suggest. In programming, a relatively small base of intensive customers is already enough to form a relevant market.

It's precisely this combination of limited adoption and high spending that deserves finance teams' attention. A company can pay for the model and, at the same time, for one or more platforms that make it easier to use. The risk isn't just in the price of each subscription, but in the overlap of tools with similar functions, contracted by different teams. Tracking spend by vendor, the number of licenses, and the concentration of usage helps distinguish a tool that's embedded in the operation from just another subscription piled onto the software stack.

The dispute, then, isn't just about which company offers the most capable model. It also involves who controls the environment in which that model turns into work. For vendors, it's a fight for space in the development process. For finance departments, it's a new spending category that has already become too big to go unnoticed.