On September 14, shares of companies related to artificial intelligence (AI) fell due to calls to slow down the development of this technology. According to Yahoo Finance, in early trading on the Asian market, shares of SoftBank, one of the major investors in OpenAI, fell by up to 13.2%. The decline also affected semiconductor manufacturers: Kioxia shares dropped by 9.8%, SK Hynix by 5.3%, Samsung Electronics and Tokyo Electron by 3.7%, and TSMC fell by 1.2%. The developer of GLM models, Z ai, lost 10.5%. This drop occurred amid a general market sell-off and the placement of new shares at a discount.
On September 13, Z ai announced it had raised $5 billion: about $2 billion came from stock sales, and the remaining $3 billion from convertible bonds. The securities were placed at 714 Hong Kong dollars, 10% below the closing price on September 11. About 60% of the net proceeds will be directed by the company towards the development of a new generation of models and their automatic learning systems, with another 15% for business expansion.
The call to slow down AI development made investors question whether investments in infrastructure related to its maintenance would be justified. Swissquote analyst Ipek Ozkardeskaya noted that if the demand for computing grows more slowly, data center revenues may not meet expectations. Meanwhile, their operators will still have to pay off debts, pay rent, and fulfill electricity purchase obligations. According to her, firms that built data centers with borrowed funds expecting rapid demand growth, as well as creditors of such projects, are particularly vulnerable.
The reason for revising expectations was an essay by Anthropic CEO Dario Amodei, supported by Sam Altman from OpenAI and Elon Musk from xAI. According to Amodei, the capabilities of models are growing faster than the means to ensure their safety. One reason he cited was the increasingly active involvement of AI in creating the next generations of systems. Another cause for concern was an incident with OpenAI and Hugging Face: as Amodei described, AI agents attacked external targets and attempted to hack the system evaluating their work. He warned that more powerful neural networks with similar behavior could cause significantly greater damage.
To mitigate risks, the head of Anthropic proposed three directions: independent audits within laboratories—external specialists should constantly have access to development processes, check compliance with safety measures, and report incidents; common rules for companies—developers should agree on safety standards and limits on the pace of model development with government support; international agreements—countries need to discuss joint measures considering the complexity of verifying their implementation. Anthropic has already committed to implementing the first point, while the others require agreements with other companies and governments.
Related articles
