Wall Street analysts react to DeepSeek's AI push as markets sell off

Investing.com -- Chinese AI startup DeepSeek shook global tech markets on Monday, fueling concerns about the US's technological edge in the AI space.

Over the weekend, excitement grew around DeepSeek's new AI model, which is cost-effective and operates on less-advanced chips, raising doubts about the lofty valuations of companies like Nvidia (NASDAQ:NVDA), a key player in the AI boom.

Nasdaq 100 futures fell by as much as 3.1%, while S&P 500 futures slid 1.9% early Monday. In Europe, tech stocks led declines, with ASML Holding NV (AS:ASML) sinking more than 10%. Meanwhile, volatility surged as VIX futures spiked.

“If DeepSeek’s innovations are adopted broadly, an argument can be made that model training costs could come down significantly even at U.S. hyperscalers, potentially raising questions about the need for 1-million XPU/GPU clusters as projected by some,” Raymond (NSE:RYMD) James analysts said in a note.

However, it appears more logical that DeepSeek's advancements are likely to push US hyperscalers to capitalize on their access to GPUs, further differentiating themselves from cost-effective alternatives.

“We believe the Stargate announcement and capex commentary from META clearly support this view,” analysts added.

Moreover, they note that declining training costs could unlock new AI use cases, driving substantial growth in inferencing, as reasoning models like R1 and OpenAI's require significantly more computational power for inference.

Founded by quant fund leader Liang Wenfeng, DeepSeek’s AI model is being compared to offerings from OpenAI and Meta (NASDAQ:META). Investor Marc Andreessen described it as “one of the most amazing and impressive breakthroughs.”

The model, which explains its reasoning in response to user prompts, quickly rose to the top of Apple’s App Store rankings after launching last week, earning praise for its transparency.

Chinese AI-related stocks surged in response. Mainland-listed Merit Interactive (SZ:300766), which had integrated DeepSeek’s model into its marketing platform, hit its daily trading limit.

Bernstein analysts said they see no reason for panic for US hyperscalers following DeepSeek's advancements, emphasizing that innovations like these are essential for AI's progression as model costs continue to rise.

They argue that efficiency gains, such as those achieved by DeepSeek, are unlikely to reduce long-term spending but instead drive higher demand for compute capacity.

The analysts believe DeepSeek's strategies are likely not unknown to other top-tier AI labs, making it improbable that such innovations are entirely unique.

“It seems like a stretch to think the innovations being deployed by DeepSeek are completely unknown by the vast number of top tier AI researchers at the world’s other numerous AI labs,” they wrote.

Bernstein points to Meta’s increased capex, the Stargate announcement, and China’s $140 billion AI spending plan as evidence that investments in AI are accelerating.

As such, they remain bullish on AI-related stocks in the US, particularly Nvidia and Broadcom (NASDAQ:AVGO), urging investors “not to buy into the doomsday scenarios” circulating online.