Google Spends the Most on AI Model Training 🚀

Key Takeaways

• While detailed information on AI model training costs is limited, the 2025 AI Index Report estimates that Google spent $192 million on Gemini 1.0 Ultra—the highest across leading models.
• These costs were calculated based on cloud compute rental prices, which can rake up millions of dollars quickly since companies rent out thousands of supercomputers that run nonstop for weeks.
• By contrast, the DeepSeek-V3 reportedly cost $6 million to train, however this cost remains disputed.

Training Costs Rise Even Higher

Today, many state-of-the-art AI models cost at least $100 million to train—a steep jump from the $670 price tab for training the groundbreaking Transformer model in 2017.

Moreover, Dario Amodei, CEO of Anthropic, said that $1 billion training costs are already underway.

For Gemini Ultra, a large chunk of the costs were for research and development staff salaries (including equity)—making up to 49% of the final cost. Meanwhile, AI accelerator chips accounted for 23% of the total cost, followed by 15% for other server components.