If AI crushes software, why is "AI leader" NVIDIA dragged down by software stocks?

The narrative of AI's impact on the software industry is triggering a round of "indiscriminate" selling, pulling down AI chip stocks represented by NVIDIA, but the logic behind the selling itself has inherent contradictions.

On Wednesday, NVIDIA's stock price fell by 3.41%, Broadcom dropped by 3.83%, Micron decreased by 10%, and SanDisk fell over 15%. Meanwhile, the selling pressure in the software sector continued, with the iShares Expanded Tech-Software Sector ETF, IGV, dropping another 1.8% on Wednesday.

On Tuesday, Anthropic launched nearly ten tools for its Claude Cowork, advancing AI applications into more scenarios such as sales and data analysis, leading the market to further internalize the narrative that "AI will defeat software" into valuation adjustments. However, the decline in software stocks triggered a broader sell-off, including the AI sector.

Bank of America analyst Vivek Arya believes that the decline in chip stocks is illogical. He pointed out that the drop in chips suggests that AI investment returns are deteriorating, while the decline in software indicates that AI adoption will become extremely widespread. These two conclusions cannot coexist, and the current market reaction is excessive.

Why the narrative of "AI defeating software" is dragging down AI chips

Vivek Arya noted that the decline in AI-related chip stocks implies a market judgment that AI investments are deteriorating to the point where ideal returns are difficult to achieve, limiting future growth potential; however, the decline in software stocks corresponds to another judgment that AI adoption will become widespread and significantly enhance productivity, to the extent that the business models of software operations and applications will be replaced.

"These two outcomes cannot happen simultaneously." Arya wrote that this trading structure is similar to the panic during the "DeepSeek shock" in January, which was ultimately proven to lack factual support. At that time, the market was concerned that DeepSeek would create competitive models at lower costs, reducing the demand for expensive high-performance chips.

AI investments show no signs of cooling, capital expenditures once far exceeded expectations

Arya believes that after the "DeepSeek shock," the market is seeing more AI spending and accelerated growth in AI tokens. In his report, he stated that cloud capital expenditures are estimated to grow by 69% year-on-year by 2025, far exceeding the initial growth expectations of 20% to 30%.

He also emphasized that even though current AI models are performing well, it may still take years to "prove value" from a productivity perspective, so he does not believe AI investment will slow down in the short term; even if significant productivity breakthroughs occur in the future, models will still need continuous improvement to maintain user engagement, and investments may not necessarily decline.

His conclusion is that, although it is impossible to determine the final shape of the software industry, the chip industry has already benefited and will continue to benefit from the AI construction cycle, adding that enterprise AI adoption is still in its early stages, and more sovereign nations are beginning to increase their AI investments Arya also pointed out that the valuation of the chip sector has already factored in concerns about a slowdown in spending or downward revisions in profit expectations, and these concerns "may not materialize."

Why Nvidia is "dragged down," what investors should look for

From a trading perspective, this round of decline resembles a "narrative-driven correlation shock": as the market simultaneously sells off software, betting on the AI substitution effect, it also sells off AI chips, betting on deteriorating AI investment returns, making core AI assets like Nvidia susceptible to being sold off during position liquidations and risk reassessments.

For Nvidia investors, what needs to be clarified in the short term is whether the current price fluctuations stem from substantive changes in AI capital expenditures and demand, or from indiscriminate de-risking across sectors triggered by the "AI defeating software" narrative.