China-based AI lab DeepSeek announced it released AI models (including R1) that took two months to build using less-powerful Nvidia Corp NVDA H800 chips, costing far less to build than LLMs today.
The exact cost of development and energy consumption of DeepSeek are not fully documented. However, BofA Securities analyst Brad Sills noted that the lower cost of running AI models benefits the general software group.
Sills noted the advancement of DeepSeek as a positive for Microsoft Corp https://www.benzinga.com/25/01/43302340/deepseeks-ai-breakthrough-microsoft-poised-to-benefit-but-oracle-faces-cloud-growth-challenges