The recent sharp decline in software stocks has now impacted the chip sector, but an analyst from Bank of America contends that this selling pressure is unfounded.
Bank of America analyst Vivek Arya stated in a report on Tuesday that the "indiscriminate" sharp volatility surrounding chip stocks "seems paradoxical, and this could be another instance of excessive selling similar to the DeepSeek event."
On Wednesday, share prices of chipmakers including NVIDIA and Broadcom fell by more than 3% and 6% respectively. NVIDIA has now declined for four consecutive days. Memory and storage stocks like Micron Technology and SanDisk, which had seen significant gains driven by AI-fueled demand, experienced drops reaching double-digit percentages.
Arya believes the extent of the selloff reflects "contradictory" viewpoints. He noted in the report that the decline in AI-related chip stocks suggests that investment in AI is "deteriorating to a point" where returns will be weak and further growth is unviable. Simultaneously, the drop in software stocks implies the market believes "AI adoption will be so pervasive and productivity-enhancing" that software-based businesses and applications are doomed to become obsolete.
"These two outcomes cannot happen concurrently," Arya said, writing that this bears similarities to concerns over DeepSeek, "which ultimately proved to be baseless." The AI startup's inference model released in January 2025 disrupted tech stock trading, as investors worried that if DeepSeek could cheaply develop competitive AI models, other companies could too, without needing as many expensive and powerful chips.
But Arya said that following that market episode, there was subsequently increased investment in AI and an acceleration in the growth of AI tokens, which are the units of information processed by AI models. He added that for 2025, cloud capital expenditure is forecast to grow 69% year-over-year, more than double the initially projected increase of 20% to 30%.
Arya stated that while current AI models are impressive, it might take several years for them to demonstrate their value from a productivity perspective. In the meantime, he does not expect AI investment to slow down and anticipates that even after significant AI-driven productivity breakthroughs, investment will continue as models will still require refinement and need to maintain user engagement.
"While we cannot predict the ultimate structure of the software industry, we believe the chip industry has already and will continue to benefit from AI infrastructure build-out," Arya commented.
Arya noted that enterprise adoption of AI is still in its early stages, and current investment is also coming from sovereign nations looking to deploy and control their own AI initiatives.
Concurrently, Arya said that valuations in the chip sector already price in caution regarding potential spending slowdowns or lowered earnings expectations, adding, "we believe such a scenario may not materialize."
Comments