By Echo Wang

NEW YORK (Reuters) -AI startup Cohere has doubled its annualized revenue since the beginning of the year, driven by growing demand for secure, customized AI tools among enterprise clients in regulated sectors, the company told Reuters.

The company has crossed an annualized revenue of $100 million as of May 2025, according to one person familiar with the matter.

A Cohere spokesperson declined to comment on the financials.

The growth followed a strategic shift in the third quarter of 2024 with a focus on private deployments tailored for customers in sectors such as finance, healthcare and government.

Chief Executive Aidan Gomez outlined the new direction in a year-end memo, emphasizing a move toward building tailored models for enterprise users over larger foundation models.

Most of its revenue stems from long-term contracts, the company said.

Roughly 85% of Cohere’s business now comes from such private deployments, the company said, with margins reaching 80%.

As part of the strategy, Cohere in January launched a ChatGPT-style application called North, aimed at helping knowledge workers with tasks such as summarizing documents. The product is currently being tested with a limited group of customers, including Royal Bank of Canada and LG.

Founded in 2019, Cohere has raised more than $900 million from investors including Nvidia (NASDAQ: NVDA ), Cisco (NASDAQ: CSCO ) and Inovia Capital. Its customers also include Fujitsu, Oracle (NYSE: ORCL ) and Notion. The company was last valued at $5.5 billion.

Cohere’s shift toward smaller, specialized models reflects a broader trend in the AI sector, as companies prioritize domain-specific tools over large, generalized systems. The move comes as AI labs report diminishing returns from scaling up model size — a strategy that once drove major breakthroughs.