AI Sector Volatility Spikes as Tariff Uncertainty Meets ROI Skepticism

The US stock market faced a sharp correction on February 23, 2026, as a perfect storm of geopolitical uncertainty and fundamental AI skepticism converged. The Nasdaq Composite led the decline, falling significantly in a single session, as investors reacted to a lack of clarity regarding new trade tariffs that threaten to disrupt the global semiconductor supply chain. This tariff fog has left analysts struggling to price in the increased costs of AI hardware, which remains the backbone of the current technological expansion. The immediate market reaction reflects a broader anxiety that the cost of building the AI future is rising just as the timeline for seeing a return on that investment is being questioned.

The tariff fog refers to the ambiguous nature of proposed trade barriers that could significantly increase the cost of importing critical components for AI servers. As the US considers new restrictions on advanced semiconductor manufacturing equipment and high-end AI chips, the entire ecosystem is on edge. For companies that rely on a globalized supply chain—spanning from design in California to fabrication in Taiwan and assembly in Southeast Asia—any disruption in trade flow translates directly to lower margins and higher end-user prices. This uncertainty is particularly damaging because it prevents long-term capital planning, forcing companies to hold higher cash reserves and delay expansion projects rather than investing in the next generation of compute.

“

The Nasdaq Composite led the decline, falling significantly in a single session, as investors reacted to a lack of clarity regarding new trade tariffs that threaten to disrupt the global semiconductor supply chain.

Simultaneously, worries over AI have transitioned from theoretical concerns about safety and ethics to practical concerns about financial sustainability. The AI CAPEX cycle has seen hundreds of billions of dollars poured into data centers, yet the revenue generated from consumer-facing AI applications has not yet reached a scale that justifies current market valuations. Large language models (LLMs) are becoming increasingly expensive to train and maintain, leading to a margin squeeze for software-as-a-service (SaaS) providers who are currently subsidizing AI features to gain market share. If the expected productivity gains do not materialize in corporate balance sheets by the end of the fiscal year, the market may witness a more profound revaluation of the entire AI stack.