Cohere's Strategic Blueprint: Sovereignty, Efficiency, and Global Reach in Generative AI

"Sovereignty means control. No one can switch it off." This assertion by Aidan Gomez, Co-Founder and CEO of Cohere, during his discussion with Bloomberg's Lynn Doan at Bloomberg Tech in London, encapsulates a core tenet of his company's strategy in the burgeoning generative AI landscape. The interview delved into Cohere's unique approach to product development, market expansion, and a contrarian view on the industry's prevailing "bigger is better" ethos, offering a sharp analysis for founders, VCs, and AI professionals navigating this dynamic sector.

Gomez articulated Cohere's distinct deployment model as a cornerstone of its "sovereignty" pitch to enterprise clients. Unlike many competitors that require customer data to be sent to their cloud for processing, Cohere empowers its clients with complete autonomy. "We send all of our software, all of our models to the customer," Gomez explained. This on-premise or air-gapped deployment means, "You can literally unplug the machine from the internet, it keeps running." This ensures unparalleled control over proprietary data and infrastructure, a critical differentiator for businesses with stringent security and compliance requirements.

This commitment to client control is intertwined with Cohere's aggressive international expansion. Gomez, whose diverse background includes Canadian, British, and Spanish heritage, noted that Cohere has always been "highly, highly international." The company has rapidly opened offices in global tech hubs like Seoul, Tokyo, Riyadh, Dubai, and Paris, alongside its London base. He views Cohere's Canadian origins as a significant asset outside the US, fostering a "friendly Canadian approach and posture" that appeals to diverse markets. This global mindset is further reflected in their product development, which prioritizes multilingualism from the outset. While many rival models are "super English-centric," Cohere ensures its models perform robustly across languages like Arabic, Korean, and Japanese, avoiding the "substantial quality drop" seen in competitors when moving beyond English.

A critical insight from Gomez challenged the prevailing narrative of infinite compute and ever-larger models. He suggested that the industry's fixation on scale, particularly in consumer-facing AGI pursuits, might be a "big mistake." Referencing the rapid evolution of models, he highlighted that the "future of LLMs belongs to those who focus on more efficient techniques, not more compute." Cohere has intentionally built its Command A model to run efficiently on just two NVIDIA H100 GPUs, a stark contrast to the dozens or even hundreds required by some other large models. This deliberate constraint, Gomez argued, is a hallmark of "good technology development" and allows Cohere to serve enterprises that lack "infinite GPUs" and are sensitive to the immense costs associated with massive compute clusters.

Related Reading

• Unpacking the AI Hierarchy: From Classic ML to Generative Intelligence
• OpenAI, SAG-AFTRA, and Cranston Confront Deepfakes in Generative AI
• AI's Dual Nature: Creature or Machine? The Battle Over Regulation

The conversation also touched upon the financial health of generative AI companies, with Gomez offering a pragmatic perspective on the "AI bubble" concerns. He noted that many competitors, particularly those chasing AGI with large consumer deployments, operate on "negative margins," essentially losing money per customer due to the exorbitant compute costs. Cohere's business model, centered on selling software licenses for private deployments, is fundamentally different. It resembles a traditional SaaS business, characterized by "very high margins." This allows Cohere to pursue a clear path to profitability "sooner" than the 2029 projections some analysts have for rivals like OpenAI.

The company's focus on delivering tangible return on investment (ROI) for enterprises, rather than speculative AGI, underscores its commercial viability. By enabling businesses to deploy and integrate AI models directly into their existing tools and workflows—from CRM to ERP systems—Cohere aims to lower barriers to adoption and accelerate real-world automation. This product-centric mission, coupled with a disciplined approach to resource allocation and a global outlook, positions Cohere as a distinctive player in the competitive generative AI landscape.