Groq and Cohere CFOs Tackle AI Profitability
Groq CFO Katie Burke and Cohere CFO Curtis Liu shared straight talk on AI profitability in a recent panel. Perplexity AI’s page rounds up their points from the event. Coverage also from The Information, Business Insider, Semafor, VentureBeat, and TechCrunch.
Groq’s Angle
Burke focused on inference costs. Groq’s custom LPUs run models fast, cutting expenses per token. She said their setup already beats GPUs on price for speed, setting up profits as customers scale up. Check Groq’s site for hardware details.
Cohere’s Angle
Liu stressed enterprise focus. Cohere prioritizes RAG tools and custom models for businesses, with contracts that build recurring revenue. He pointed to improving gross margins from better utilization. More at Cohere’s site.
Common Ground
- Both see inference as the money-maker over training.
- Scale solves compute burn.
- Profitability comes 12-24 months out for leaders.
The discussion shows hardware and software paths converge on cost control. No hype—just real math on revenue vs. chips.
