From Financial Modeling Prep: 2025-01-28 05:42:09

Analysts at Cantor Fitzgerald dismiss concerns over “peak GPU spending” with the release of DeepSeek V3, highlighting the demand for AI compute power benefiting Nvidia and the AI ecosystem. DeepSeek V3, compared to GPT-4 and Llama 3.1, shows potential for AGI advancements, emphasizing the need for greater computational capacity.

The Jevons Paradox in AI predicts increased efficiency will drive greater usage, making AI technologies more prevalent and boosting demand for GPUs. Open-source frameworks like PyTorch and Llama enhance cost-efficiency in training and deploying advanced models, strengthening the outlook for GPU manufacturers like Nvidia.

Cantor questions DeepSeek V3’s reported $5.57 million training cost using 10,000 A100 GPUs instead of rumored 50,000 Hopper GPUs. Concerns arise over discrepancies and a possible “GPU embargo,” prompting analysts to seek clarification on the situation.

Cantor’s report indicates the AI boom continues, with AGI advancements fueling demand for new computing infrastructure. This aligns with expectations of increased investment in AI-focused technologies and the expansion of GPU production to meet growing demands.



Read more at Financial Modeling Prep:: DeepSeek V3: Cantor Fitzgerald Sees Bullish Implic…