Citibank analysts said that although DeepSeek's success has raised questions about the cost of AI development, AI spending should continue to grow strongly for now, and pointed out that the company's model is based on the use of cloud service providers, and the technology is still in its infancy. Morgan Stanley is also still optimistic about chipmakers, but it has cut some of its target prices.
Analyst comments:
Citi analyst Christopher Danely said that although AI spending growth does not seem to be slowing down, “eventually we need to see some kind of return on investment (ROI), or else spending may slow rapidly.”
“Whenever AI spending slows drastically, Broadcom will be most affected, given its 24% AI exposure and 70% multiplier expansion,” Danely wrote.
Note: Broadcom's stock price plummeted 17% on Monday, and the market value evaporated by nearly 200 billion dollars
Citi said, “Given Micron's AI exposure is 8%, the company will also be affected,” and “AMD will also be affected, but less so, because its MI300 product is ideal for the open AI system used by DeepSeek, which may be a good thing for AMD.”
Morgan Stanley analyst Joseph Moore said, “We were surprised by the strong reaction to the DeepSeek release.”
J.P. Morgan analyst Harlan Sur said, “As DeepSeek has shown, increased AI efficiency should stimulate greater demand and drive further innovation.”
“Innovation in large-scale language models is driving innovation in new AI model architectures,” Sur wrote.
Rosenblatt said the emergence of another competitive large-scale language model is good news for Ambarella, CEVA, Qualcomm, and Synaptics' “network edge AI processors.”
American hyperscale cloud service providers (hyperscalers) will learn from these models and “don't think they will cut capital expenses on data centers, GPUs, fiber, and Ethernet as a result.”
Comment(0)
Reason For Report