AI Pulse · Business Models & Winners

Z.AI, the Chinese AI company behind the GLM coding models, went from roughly $100 million in annual recurring revenue in January to $1 billion by June. That’s five months to cover ground that took Anthropic fifteen months, according to Bloomberg’s reporting. The driver is mostly GLM-4.6, a coding model that’s found traction with developers, plus about 100,000 monthly API users outside China.

The speed matters because it shows coding assistance is a category where customers will pay real money right now, not in some vague future quarter. Developers need help shipping faster, and a model that handles code well can slot straight into existing workflows without a long sales cycle. Z.AI’s growth also highlights how quickly a well-targeted product can scale when the use case is concrete and the buyer has budget authority.

But the same report notes Q1 net losses jumped 60% alongside that revenue climb. Fast top-line growth and widening losses often travel together in AI, where inference costs, talent, and compute all burn cash. It’s a reminder that ARR velocity tells you about demand, not whether the business works at scale. If you’re evaluating an AI vendor or building something in-house, the unit economics question, how much does each dollar of revenue actually cost, matters as much as the headline number.

For operators thinking about how to track performance and cost across your own AI tools, this is exactly the kind of metric you want visible in one place. We build that into the Omni Command Centre so you can see what’s working and what’s just expensive. The gap between revenue and profit in AI is real, and it shows up fastest when you’re not watching both at once.