The artificial intelligence sector has just received a warning signal from its own command center. Sector-linked stocks fell Monday on Wall Street, Europe, and Asia after executives from some of the leading AI companies raised the need to moderate the pace of development for increasingly capable models.
The stock market movement exposes an issue that goes far beyond technological security: how much capital the infrastructure needed to build the next generation of AI can continue to absorb.
Dario Amodei, CEO of Anthropic, called for a reduction in the pace of advancement in the face of risks he considers potentially serious. Sam Altman of OpenAI and Elon Musk of xAI expressed their agreement with the need to act against these risks. Altman added that OpenAI will not pursue an IPO this year.
The statements arrived as markets began to question how long the extraordinary investment cycle that turned AI into one of the main drivers of the stock markets can be maintained.
Chipmakers take the first hit
The effect was particularly strong in semiconductors. The Philadelphia Semiconductor Index fell 5.2%, while Nvidia dropped 3%, AMD 4.5%, and Micron 5.4%. In Europe, the technology sector declined 2.2%, with ASML losing around 6%; in Asia, SoftBank fell more than 10%, and manufacturers such as TSMC and SK Hynix also retreated. The Nasdaq 100 lost as much as 1.2% during the session before paring some of the losses.
The reaction is technically significant because semiconductors constitute one of the first physical links in AI expansion. Every increase in training and inference capacity requires processors, memory, high-speed networks, and cooling systems, in addition to data centers capable of supplying enormous amounts of energy.
A slowdown in model development could modify expectations for the future utilization of all that infrastructure.
The financial problem amplifies the risk. AI growth has relied on massive investments and, increasingly, on debt and financing mechanisms between companies within the ecosystem itself.
Reuters notes that global AI-related spending could approach $800 billion during 2026 and exceed $1 trillion in 2027. Therefore, a reduction in the pace of expansion would not only affect the companies that train models: it could impact chipmakers, equipment suppliers, data centers, and energy companies.
Security and technological capacity collide
The concern does not arise solely from abstract statements. Anthropic recently published a report on operations in which malicious actors used Claude for activities related to cyber-operations, surveillance, fraud, influence, and weapons development.
The company reported that it identified and disrupted these operations, using the incidents to reinforce its protection mechanisms. The very evolution of AI agents is changing the nature of the risk.
Anthropic explains that its systems are acquiring the capacity to execute tasks for longer periods and operate with access to digital tools and systems. The company recognizes that as the capacity of agents increases, so does the so-called blast radius: the potential impact of a failure or misuse.
Companies need to develop more capable systems to compete, but that same capacity increases the need for controls, testing, and containment mechanisms. Slowing down too much could mean losing a technological advantage; moving too fast could increase operational and regulatory risks.
China adds pressure on Artificial Intelligence
The security debate is also occurring within an international technological competition. OpenAI and Anthropic face growing pressure from Chinese models seeking to offer comparable capabilities at lower costs.
Among them are proposals such as Kimi K3, Qwen, and DeepSeek models. This pressure could reduce the prices of AI services and make it difficult for U.S. companies to recover the massive investments made in infrastructure.
The tension explains part of the difficulty in accepting a coordinated slowdown. If one company reduces its development pace while its competitors continue to advance, it may lose its technological and commercial position.
SOURCE: https://www.reuters.com/
PHOTO: Canva
