While OpenAI built ChatGPT and Anthropic created Claude, a Toronto startup took a different path: enterprise-only AI with no consumer chatbot. That bet paid off. Cohere is now Canada's most valuable AI company at $7 billion, with Oracle, McKinsey, and RBC among its customers.
What Is Cohere?
Cohere is a Toronto-based artificial intelligence company that builds large language models (LLMs) designed exclusively for enterprise use. Unlike consumer-focused competitors, Cohere's models can be deployed on any cloud, on-premise, or in air-gapped environments.
The company was founded by three researchers with deep connections to the transformer architecture that powers all modern AI. CEO Aidan Gomez co-authored the original "Attention Is All You Need" paper, while President Nick Frosst was Geoffrey Hinton's first hire at Google Brain Toronto.
What Products Does Cohere Offer?
Cohere provides a complete platform for enterprise AI, from text generation to search optimization to autonomous agents.
Core Product Suite
Command R and Command R+
Command R+ is Cohere's flagship model—a 104-billion-parameter LLM optimized for enterprise use cases. Key capabilities include:
- RAG optimization: Built-in citation support to reduce hallucinations
- Multilingual: Native support for English, French, Spanish, Italian, German, Portuguese, Japanese, Korean, Arabic, and Chinese
- Tool use: Can connect to external APIs and automate multi-step workflows
- Long context: 128,000 token context window
The smaller Command R7B offers similar capabilities in a 7-billion-parameter package, making it suitable for latency-sensitive applications while still supporting the full 128K context window.
Embed and Rerank
Cohere's retrieval products power enterprise search and RAG systems:
Embed converts text and images into vector embeddings for semantic search. It works across 100+ languages and can match queries to documents even when they're in different languages.
Rerank re-orders search results to surface the most relevant documents. The latest version (Rerank 3) supports 4,000-token context and can process semi-structured data like emails, invoices, JSON, code, and tables.
Together, these products enable a two-stage retrieval architecture: first, dense retrieval via Embed finds candidate documents, then Rerank boosts the most relevant results to the top.
North: The AI Agents Platform
In August 2025, Cohere launched North—a low-code platform for building and deploying AI agents across the enterprise.
North competes directly with Microsoft's autonomous agents, Google's Vertex AI agents, and Salesforce's Agentforce. The key differentiator is deployment flexibility—North can run entirely within an enterprise's own infrastructure.
"The hype is 100% justified... These things are going to transform productivity, and they're coming." — Aidan Gomez, Cohere CEO, on AI agents (August 2024)
How Is Cohere Different from OpenAI and Anthropic?
The three leading AI companies have taken fundamentally different approaches to the market.
Strategic Positioning
The Privacy Advantage
Cohere's enterprise focus addresses what many consider the biggest barrier to AI adoption: data privacy.
- No data training: Cohere does not use customer data to train models (unless explicitly agreed for fine-tuning)
- Firewall deployment: Models can run entirely within customer infrastructure
- Cloud-agnostic: Not locked into any single cloud provider
- Compliance: SOC 2 Type II, GDPR, HIPAA certified
"We have a differentiated deployment model, we have very high margins similar to a traditional SaaS business, whereas many others in the industry are losing money per customer." — Aidan Gomez, Bloomberg Tech 2025
While OpenAI's close alignment with Azure and Anthropic's partnership with AWS create vendor lock-in concerns, Cohere's models run on AWS, Azure, Google Cloud, Oracle Cloud, or customer data centers.
Who Are Cohere's Major Customers?
Cohere has built an impressive roster of enterprise customers across industries.
Key Partnerships
Industry Verticals
Financial Services: RBC is co-developing "North for Banking" as an AI assistant platform tailored to financial services. Cohere's compliance certifications (SOC 2, GDPR, HIPAA) make it suitable for regulated industries.
Healthcare: Ensemble Health Partners launched an end-to-end agentic AI platform for revenue cycle management using Cohere technology.
Telecom: Saudi Telecom (STC) is building "North for Telecom" to automate operations with AI agents.
Government: In December 2024, Cohere received $240 million from the Canadian government as part of the Canadian Sovereign AI Compute Strategy. The company also announced an alliance to deploy AI tools across Canadian and U.K. governments.
How Much Has Cohere Raised?
Cohere has raised approximately $1.6 billion across multiple funding rounds.
Complete Funding History
Revenue Growth
Cohere's revenue has grown dramatically:
Approximately 85% of revenue comes from private deployments, with large enterprise customers signing multi-year contracts.
What Is Cohere's Technology Stack?
Model Architecture
All of Cohere's models are built on the transformer architecture—fitting, given that CEO Aidan Gomez co-authored the original transformer paper.
Training Infrastructure
Cohere trains and deploys models on Oracle Cloud Infrastructure (OCI), using Kubernetes Engine (OKE) and GPUs. This partnership ensures enterprise-grade reliability and global availability.
Deployment Options
What's Next for Cohere?
Near-Term Priorities
- IPO Preparation: CEO Aidan Gomez has indicated Cohere could go public "soon"
- North Platform Expansion: Scaling the AI agents business with industry-specific versions
- Government Contracts: Expanding deployments across Canadian and U.K. governments
- Profitability: Targeting profitability before 2029
The Efficiency Thesis
While competitors chase ever-larger models, Cohere has bet on efficiency:
"The future of LLMs belongs to those who focus on more efficient techniques, not more compute." — Aidan Gomez, Bloomberg Tech 2025
This philosophy—building practical, deployable AI rather than benchmark-chasing—continues to guide Cohere's strategy and resonates with enterprise customers focused on ROI.
Frequently Asked Questions
What does Cohere do?
Cohere builds large language models (LLMs) for enterprise use. Unlike OpenAI or Anthropic, Cohere doesn't have a consumer chatbot—it focuses entirely on B2B customers who need AI that can be deployed privately, on any cloud, or on-premise.
How is Cohere different from ChatGPT?
ChatGPT is a consumer product from OpenAI. Cohere doesn't have a consumer chatbot. Instead, Cohere provides enterprise APIs and deployment options that let companies run AI models within their own infrastructure, maintaining data privacy and sovereignty. Cohere also offers multi-cloud deployment (AWS, Azure, GCP, Oracle) while ChatGPT/OpenAI is primarily Azure-aligned.
What is Cohere's Command R+?
Command R+ is Cohere's flagship language model—a 104-billion-parameter LLM optimized for enterprise use cases. It supports 128K context, 10 languages, retrieval-augmented generation (RAG), and multi-step tool use. It's designed for complex reasoning tasks and workflow automation.
What is Cohere North?
North is Cohere's AI agents platform, launched in August 2025. It allows enterprises to build and deploy autonomous AI agents that can complete multi-step tasks, connect to workplace tools (Gmail, Slack, Salesforce), and run entirely within the customer's infrastructure for maximum security.
Who are Cohere's biggest customers?
Major customers include Oracle, McKinsey, RBC, Dell, LG, Notion, Salesforce, and Palantir. Oracle has embedded Cohere's models into Oracle Fusion Cloud and NetSuite, reaching over 400,000 customers. RBC is co-developing a banking-specific version of the North platform.
How much is Cohere worth?
As of September 2025, Cohere is valued at $7 billion. The company has raised approximately $1.6 billion in funding and generates around $150 million in annual recurring revenue.
Is Cohere profitable?
Not yet, but CEO Aidan Gomez has stated Cohere operates with "very high margins similar to a traditional SaaS business" and expects profitability before 2029. Unlike competitors who lose money on inference, Cohere's enterprise deployment model generates sustainable unit economics.
Can Cohere models run on-premise?
Yes. Cohere is one of the few major AI companies offering true on-premise deployment. Models can run in customer data centers, air-gapped environments, or private clouds—critical for regulated industries and government customers.
Related Reading
The Cohere Story
- Aidan Gomez: Cohere Founder Complete Guide — How a 20-year-old intern co-authored the transformer paper and built a $7B company
- Geoffrey Hinton: Godfather of AI — The Nobel laureate who invested in Cohere and hired Nick Frosst
Canadian AI Ecosystem
- Canadian AI Unicorns: Complete Guide — Cohere and the other $1B+ Canadian AI companies
- Toronto: Canada's AI Enterprise Capital — Why enterprise AI companies choose Toronto
- The AI Triangle: Canada's Three Superpowers — Toronto, Montreal, and Vancouver's AI collaboration
The Research Pipeline
- Vector Institute Complete Guide — Toronto's AI research and talent hub
- Yoshua Bengio: Deep Learning Pioneer — The Turing Award winner building AI safety tools
Building in Canada
- AI Founder Resources Canada — Programs and funding for AI founders
- Canadian AI Accelerators & Incubators — Where to launch your AI startup
Join AGI House Canada to connect with Toronto's AI founder community. Scan the QR code to join our WhatsApp group or subscribe to our newsletter.
Data current as of January 2026. Revenue estimates from Sacra. Funding data from Cohere announcements, Crunchbase, and TechCrunch. Product information from Cohere documentation.
