This graphic is part of AI Week, our Terzo-sponsored series exploring how artificial intelligence is reshaping business, technology, and society.
Key Takeaways
- Compute is the biggest cost center for these AI companies, accounting for 70% of Anthropic’s spending, 57% of Minimax’s, and 61% of Z.ai’s.
- AI labs are spending heavily not just to train models, but also to run inference at scale—Anthropic spent $4.1B on R&D compute and $2.7B on inference compute, versus $0.14B and $0.04B for Minimax, and $0.18B and $0.01B for Z.ai.
- The data highlights how expensive it is to compete in AI, with total spending reaching $9.7B for Anthropic, compared with roughly $0.32B for Minimax and $0.31B for Z.ai.
