Cohere — Enterprise AI from Canada

The Company That Chose Enterprises Over Consumers

While AI companies race for individual users, Cohere chose a different path from day one: serving companies and governments that prioritize privacy and data sovereignty. This choice earned it a distinct share in the enterprise market, with a $7 billion valuation and annual revenue exceeding $240 million as of February 2026.

Founding Story: From “Attention Is All You Need” to Enterprise Independence

Cohere was founded in 2019 in Toronto by three University of Toronto alumni, most notably Aidan Gomez (CEO), who was one of the co-authors of the seminal 2017 research paper “Attention Is All You Need” — the paper that established the foundations of Transformer architecture and launched the modern generative AI revolution. Joining him were Ivan Zhang and Nick Frosst.

The team launched with a clear philosophy: AI for enterprises must be private, secure, and customizable, not a general-purpose tool exposed on the internet. This stance, which seemed conservative in 2019, became a genuine competitive advantage in 2025-2026 after governments and financial companies began viewing shared models with millions of users with increasing skepticism.

Main Products and Models

Command Family

Generative language models customized for enterprise tasks like advanced chat, document analysis, and content creation.

• Command A: Launched in March 2025, matches GPT-4o performance on many tasks, with 75% faster response speed.
• Command R+: The flagship model for complex tasks requiring deep reasoning and context up to 128,000 tokens.
• Command R7B: A smaller, more efficient model suitable for enterprises needing speed at lower cost.

Embed

Multilingual text-to-embedding models excelling in semantic search understanding. The latest version supports processing documents up to 200 pages and handles multimodal inputs.

Rerank

A specialized model for reordering search results by semantic relevance rather than keyword matching. Significantly improves enterprise search system accuracy.

North

An enterprise productivity platform launched by Cohere in August 2025, operating with AI agents capable of syncing data between enterprise applications and automating tasks. Chief AI Officer Joelle Pineau (formerly leading Meta’s FAIR lab) announced that developing North is her top priority.

Aya Vision (Research)

A multilingual, multimodal model launched by Cohere Labs in March 2025, capable of image description, translation, and summarization. Available free for research purposes.

Achievements and Numbers

• Total Funding: $1.54 billion across 7 funding rounds.
• Latest Round (September 2025): $600 million at $6.8 billion valuation, then extended by an additional $100 million, bringing valuation to $7 billion.
• Key Investors: Radical Ventures, Inovia Capital, NVIDIA, AMD Ventures, Salesforce Ventures, Healthcare of Ontario Pension Plan.
• Revenue: $100 million annually in May 2025, rising to $150 million by end of October 2025, then $240 million in February 2026.
• Employees: 842 as of February 2026.
• November 2026 (Potential): Merger discussions with German Aleph Alpha — could create a European-Canadian bloc independent of major American tech companies.

Enterprise Partnerships

• Oracle: Integrating Cohere technology into Oracle Fusion Cloud and Oracle NetSuite.
• SAP: Integrating Cohere models into SAP Business Suite and SAP AI Core.
• McKinsey: Partnership to help enterprises integrate AI into their operations.
• Bell Canada: Providing AI services to Canadian governments and enterprises.
• RBC: Launching North for Banking — a secure AI platform for banking services.
• AMD: The only major AI provider to make models compatible with AMD Instinct GPUs, with AMD also using Cohere models internally.

Competition and Challenges

The essential difference between Cohere and its larger competitors like OpenAI and Anthropic lies in one point: data sovereignty. While those companies rely on shared cloud infrastructure, Cohere enables deployment of its models on-premises, in private clouds (VPC), or even in completely air-gapped configurations.

This option makes it a primary candidate for:

• Financial institutions and banks governed by strict data protection rules.
• Governments wanting to deploy AI without sending data to foreign servers.
• Defense and healthcare sectors.

The biggest challenge Cohere faces is the large gap between it and OpenAI (estimated at $500 billion) and Anthropic ($183 billion) in terms of valuation and resources. However, it is more diverse in its investors, not relying on a single funder the way its competitors depend on Microsoft or Amazon.

Future Vision 2026–2027

Geographic Expansion: North America, Europe, Asia-Pacific, Middle East, and Africa. Opened offices in Paris, South Korea, and Montreal.

Potential Merger with Aleph Alpha: In April 2026, it was announced that both companies are in advanced merger discussions supported by the German government. This merger could produce a European-Canadian competitor with real weight in the digital sovereignty market.

Potential IPO: CEO Aidan Gomez announced interest in a London listing. The appointment of François Chadwick as the company’s first CFO (who managed Uber’s IPO) reinforces the seriousness of this direction.

Analytical Conclusion

Cohere chose a narrower but deeper battle: enterprises that cannot or do not want to hand over their data to others. This space is real and growing, and the numbers support it ($240 million annual revenue with growth exceeding 200% in a year).

The three main risks ahead: that OpenAI and Anthropic decide to seriously expand into the on-premises enterprise market, that privacy technologies improve in ways that reduce its competitive advantage, and that the path to profitability is longer than investors hope. But to date, the strategy is working, and the market is rewarding it.