Toronto, not San Francisco
Cohere was founded in Toronto in 2019 by Aidan Gomez, Ivan Zhang and Nick Frosst, three University of Toronto connections who had worked at Google Brain; Gomez was a coauthor of the 2017 Transformer paper that underpins the current generation of language models. Gomez remains CEO. The leadership team also includes Joelle Pineau, previously Meta's head of AI research, as Chief AI Officer, and Phil Blunsom as Chief Technology Officer. The company keeps a base in Toronto alongside offices in San Francisco, New York, London, Montreal, Paris, Seoul and Berlin. [1]
Its product line runs through the Command family of language models, now at Command A, alongside Embed and Rerank for search and retrieval, Aya for multilingual work, and North, an agentic platform for workplace tasks launched in 2025. Cohere raised close to $1 billion across four rounds between 2021 and 2024, then a $500 million round in 2025 that valued it at $6.8 billion, followed by a further $100 million later that year that pushed the valuation to roughly $7 billion. The company reported $240 million in annualized revenue as of February 2026. [1][2]
Selling to banks instead of consumers
Cohere has never shipped a consumer chatbot to compete with ChatGPT or Gemini. Its pitch is aimed at regulated industries, banks, insurers, governments and defense, where a customer wants a model it can deploy inside its own environment rather than call over the public internet, with named customers and partners including RBC, Oracle, Salesforce, SAP, McKinsey, Thales Group, Saab and the governments of Canada and the UK. That is a narrower market than the consumer-scale labs are chasing, and Cohere's relevance to AI history sits mostly in demonstrating that a frontier-adjacent model company can build a business there instead. [1]
Atlas interpretation: The gap between Cohere's roughly $7 billion valuation and the valuations OpenAI, Anthropic and Google command is itself informative: it marks out enterprise, on-premises deployment to regulated buyers as a smaller and more defensible market than general-purpose consumer AI, not a larger one. [1][2]
One of sixteen signatures in Seoul
Cohere is one of the sixteen original signatories of the Frontier AI Safety Commitments signed at the AI Seoul Summit on May 21, 2024, alongside OpenAI, Anthropic, Google, Meta, Microsoft, Amazon and others. The commitments are voluntary: signatories agreed to publish safety frameworks defining risk thresholds and to withhold or pull back models that cross them, with no external enforcement mechanism attached. Cohere had earlier signed the White House's voluntary AI commitments and Canada's code of conduct for generative AI in September 2023, so Seoul extended a pattern rather than starting one. [3]
Atlas interpretation: Cohere's presence on that list is a reminder that the roster of companies whose models could plausibly cross a stated risk threshold is broader than the two or three labs that dominate public attention, even if Cohere's enterprise-only distribution means fewer people ever interact with its models directly. [3]
Sources
- About Our Company | Cohere
Cohere
- Cohere
Wikipedia · Sep 9, 2026
- Frontier AI Safety Commitments, AI Seoul Summit 2024
Department for Science, Innovation and Technology · May 21, 2024