20VC with Harry Stebbings• Sep 1, 2025• 1:14:38Interview

Cohere Founder, Nick Frosst: How To Compete with OpenAI & Anthropic, and Sam Altman’s AI Disservice

From 20VC with Harry Stebbings

Nick Frosst•Founder, Cohere

Executive Summary

Cohere's co-founder presents a pragmatic, enterprise-focused vision for AI, contrasting with the industry's prevalent AGI hype, which he views as a damaging form of misinformation.
The core bottleneck in advancing LLMs is high-quality data, not algorithms or compute, and the Transformer architecture has remained largely unchanged since its 2017 invention.
Cohere differentiates through capital efficiency, spending orders of magnitude less than competitors, and by training specialized models for enterprise use cases with synthetic business data.
The future of knowledge work will be dominated by language-based interfaces, where users automate complex, multi-step tasks by giving natural language commands to AI agents.

12 quotes

Concerns Raised

The public hype around AGI is the most damaging and confusing form of misinformation in the industry.
High-quality data remains the primary bottleneck for improving large language model utility.
The impact of AI on income inequality is a significant risk that depends heavily on future labor policy.

Opportunities Identified

Singularly focusing on the enterprise market for large language models.
Developing sovereign AI capabilities for nations seeking non-US technology partners.
Automating complex, multi-step tasks for knowledge workers through language-based AI agents.
Building highly efficient models that offer a better performance-to-cost ratio for enterprise deployment.

Key Themes

Pragmatism Over Hype

The speaker argues that the public discourse around Artificial General Intelligence (AGI), particularly statements from figures like Sam Altman, is a disservice to the technology. He asserts that current LLMs, based on the Transformer architecture, are not on a path to AGI and that focusing on practical, current applications is more productive.

This provides a grounded, contrarian perspective in a hype-driven industry, emphasizing real-world business value over speculative, existential narratives, which can help enterprises set realistic AI strategies.

The Enterprise-First AI Strategy

Cohere is singularly focused on the enterprise market. This focus dictates their entire strategy, from using synthetic business data for training to optimizing models like Command R+ to run efficiently on common enterprise hardware (e.g., two GPUs).

This highlights a distinct go-to-market strategy in the foundational model space, prioritizing total cost of ownership, data privacy, and specific business tool-use capabilities over chasing consumer-facing chatbot dominance.

Data as the Primary Bottleneck

Despite the industry's focus on compute scaling laws, the speaker identifies high-quality data as the most significant constraint to improving LLM utility. While synthetic data helps, the process still relies on a foundation of real-world data, making data acquisition and curation a critical competitive vector.

This insight shifts the strategic focus for businesses looking to leverage AI from simply securing more GPUs to developing robust data strategies, as model performance is fundamentally limited by the quality of its training inputs.

The Future of Human-Computer Interaction

Within the next decade, the primary way knowledge workers will interact with computers will be through natural language. The skill of 'prompting' will diminish as models become better attuned to user intent, enabling reliable automation of complex tasks like filing expense reports through simple commands.

This vision outlines the next paradigm of user interfaces and software, suggesting that companies should prepare for a shift away from graphical interfaces towards conversational, agent-based systems for core business processes.

Capital Efficiency and Geopolitical Diversification

Cohere has built its foundational models with orders of magnitude less capital than competitors, emphasizing efficiency. Furthermore, its Canadian nationality is presented as a business asset, appealing to international clients seeking AI partners outside the US tech ecosystem and supporting the idea of sovereign AI.

This demonstrates an alternative, more sustainable path to building a major AI company and introduces geopolitical considerations into AI procurement, suggesting that nationality can be a key differentiator.

Get started free

Topics

AI Strategy Large Language Models (LLMs)Enterprise AI Cohere Transformer Architecture AGI (Artificial General Intelligence)AI Hype Data Bottleneck Synthetic Data Model Training Compute Efficiency AI Business Models Open Source vs. Closed Source AI Future of Work Sovereign AI Sam Altman Google Brain Prompt Engineering

Processed Mar 29, 2026 yt-dlp + mlx-whisper + Gemini