Mistral board member and a16z VC Anjney Midha says DeepSeek won’t stop AI’s GPU hunger
Anjney Midha discusses DeepSeek's AI advancements and ongoing demand for GPUs.

Anjney Midha, a board member of Mistral and a partner at Andreessen Horowitz, highlights that DeepSeek's R1 model, despite its efficiency, still demands significant GPU resources. He mentions that Mistral, though overshadowed in funding by the likes of OpenAI, maintains competitiveness through its open-source strategy, which grants access to collaborative technical labor and computing resources.
Midha's role in a16z’s Oxygen GPU sharing program underscores the relentless demand for GPUs in AI, with the program being overbooked due to the necessity for both model training and product operations. This demand extends to companies like OpenAI, which continue major partnerships to secure AI data center infrastructure, reflecting the broader appetite for GPU access.
The emergence of DeepSeek has prompted considerations of AI as foundational infrastructure akin to electricity or the internet. Midha urges Western nations to adopt Western AI models to preserve ethical and operational independence from models controlled by China. Nevertheless, some view this concern skeptically, noting that American firms already offer secure DeepSeek options globally.