Hopper58 schreef op 27 januari 2025 17:53:
Why These Five-Star Analysts Argue that DeepSeek’s LLM Will Boost Demand for GPUs
Vince Condarcuri
Jan 27, 2025, 05:28 PM
The release of DeepSeek’s V3 large language model (LLM), which was built for only $5.7 million, has caused investors to worry, which has led to a decline in stocks related to artificial intelligence. This includes Nvidia NVDA -15.91% ? , Broadcom AVGO -16.52% ? , Advanced Micro Devices AMD -6.15% ? , and ASML Holding ASML -6.80% ? . However, five-star analysts at Cantor and UBS disagree with the negative sentiment by arguing that more efficient LLMs will actually boost demand for companies that produce high-end graphic processing units (GPUs) and those that build data centers.
Cantor analysts, led by C.J. Muse, believe that the increased efficiency of DeepSeek’s V3 model will lead to increased demand for computing power due to The Jevons Paradox, which states that increased efficiency leads to increased usage. They argue that the AI industry will want more computing power, not less. In addition, UBS analysts, led by Timothy Arcuri, share a similar view and believe that compute demand for AI will not fade despite DeepSeek’s efficiency gains.
Interestingly, the analysts also questioned the reported cost of developing DeepSeek’s V3 model, which is significantly lower than expected. Indeed, they suggest that the actual cost may be higher and that DeepSeek may have used more advanced GPUs than initially stated. Despite these questions, the analysts agree that DeepSeek’s advances bring us closer to achieving artificial general intelligence, which will increase investment in large-scale computing clusters and benefit the AI industry as a whole.
www.tipranks.com/news/why-these-five-...