(Bloomberg) — Artificial intelligence startup Cohere Inc. is now one of the world’s most valuable artificial intelligence companies, and one of the largest startups in Canada — but unlike some of its Silicon Valley competitors, it’s not particularly flashy.
Most Read from Bloomberg
In a new funding round, Cohere was valued at $5.5 billion, vaulting it to the upper echelons of global startups. It landed there without a consumer app that writes poems, draws pictures or helps with homework.
Instead, Toronto-based Cohere makes large language models — software trained on massive swaths of the internet to analyze and generate text — and customizes them for businesses. Its software has attracted hundreds of customers such as Notion Labs Inc. and Oracle Inc. (also an investor), which use the startup’s technology to do things like help write website copy, communicate with users and add generative AI to their own products.
Cohere has also attracted investors. The company has raised $500 million in a Series D funding, it plans to announce on Monday. The round was led by Canadian pension investment manager PSP Investments, alongside a syndicate of additional new backers including investors at Cisco Systems Inc., Japan’s Fujitsu, chipmaker Advanced Micro Devices Inc.’s AMD Ventures and Canada’s export credit agency EDC.
The fresh financing more than doubles the startup’s valuation from last year, when Cohere raised $270 million in a round led by Montreal-based Inovia Capital, and brings its total cash haul to $970 million. The round has also coincided with an increasingly competitive landscape for venture funding, even in the closely watched world of AI. Reuters previously reported some details of the deal.
Cohere is one of only a handful of startups building massive large language models from scratch, partly because the technology is extremely expensive and difficult to construct. Competitors include the likes of OpenAI, Anthropic and Google. OpenAI in particular has said its goals are wildly ambitious — attempting to build artificial general intelligence, or AGI, meaning AI software capable of performing as well as (or better than) humans at most tasks.
Cohere is instead pursuing the immediately practical goal of making software to help companies run more efficiently. “We’re not out there chasing AGI,” said Nick Frosst, one of the company’s three co-founders. “We’re trying to make models that can be efficiently run in an enterprise to solve real problems.”
Started in 2019, Cohere is led by co-founder Aidan Gomez, who is a genuine celebrity in the world of artificial intelligence. Gomez is one of the authors of the seminal research paper “Attention Is All You Need,” which led to advances in the ways computers analyze and generate text. Gomez, Frosst and co-founder Ivan Zhang, have built the company rapidly in the years since. This spring, they rolled out Cohere’s new model, Command R+, the company’s most powerful so far. Cohere says it’s intended to compete against rivals like OpenAI, while costing less.
At the end of March, Cohere was generating $35 million in annualized revenue, up from $13 million at the end of 2023, according to a person familiar with the matter who asked not be identified because the information is private. The company, which started the year with roughly 250 employees, plans to double its headcount this year.
The capabilities of large language models have changed quickly over the past four years, and public interest in chatbots that run on such software — which can capably mimic human conversations — skyrocketed since late 2022 with the launch of OpenAI’s ChatGPT. Figuring out how to make the technology useful and staying ahead of the curve as it evolves has been a major effort for the company, Frosst said.
Today, Cohere has customers across a wide range of industries. They include banks, tech companies and retailers. One luxury consumer brand is using a virtual shopping tool Cohere built to help workers suggest products to customers. Toronto-Dominion Bank, a new customer, will use Cohere’s AI for tasks such as answering questions based on financial documents, Frosst said.
“My favorite use cases of this technology are the are ones that power things that nobody wants to do,” Frosst said. For example, a startup called Borderless AI uses Cohere’s models to answer questions related to the intricacies of employment law around the world in multiple languages.
Cohere’s models can be used across 10 languages, including English, Spanish, Chinese, Arabic and Japanese, and its models can cite sources in answers as well.
Guillermo Freire, who heads the mid-market group at EDC, said the startup’s ability to operate across many languages was one of the things that interested the Canadian government agency. Freire hopes EDC’s investment will help the homegrown company expand internationally, but remain based in the country. Cohere has now grown to include offices in hubs like San Francisco and London, but says it doesn’t plan to leave the city where it started. “Toronto’s been a great place to build a global company,” Frosst said.