Who might actually welcome a deceleration in AI development? Investors in the software sector, that's who. Enterprise software companies including Salesforce, Shopify, and ServiceNow all saw their share prices climb on Monday. ServiceNow surged 7%, while Salesforce gained close to 5%. For a long time, market consensus has held that software firms face significant disruption from artificial intelligence.

Meanwhile, shares of neocloud providers—cloud services specializing in AI computing rentals—and chip companies experienced declines of a similar magnitude. If the pace of AI development truly slows, these companies could face severe consequences. Wall Street's reaction on that day partially illustrates who stands to benefit and who stands to lose should the AI industry enter a real downturn. However, this is likely just a one-day emotional response from investors.

While figures like Anthropic's Dario Amodei have called for deliberate slowing of frontier AI development, achieving a coordinated and synchronized slowdown across the industry is nearly impossible. Both the Trump administration and Chinese authorities have dismissed AI safety concerns as a "scam" or "alarmism"—a rare point of agreement between the two sides. At the same time, tech giant Meta has remained unusually quiet, with only its affiliate, Alexander Wang, posting related commentary on X. Admittedly, as many have pointed out, Anthropic and OpenAI could voluntarily decelerate their own model iteration paces if they chose to, and OpenAI has previously published blog posts about managing model development speed.

But even if both leading firms like OpenAI and Anthropic proactively slow down, the effect would be minimal if Chinese model research institutions and companies like Meta continue at full throttle. It's similar to the Manhattan Project scenario where some scientists called for a pause in research while other participants simply pushed the proponents of slowing down aside. After reviewing the METR investigation report on the OpenAI‑Hugging Face incident, it's hard not to feel concerned about the capabilities and autonomous behaviors demonstrated by AI agents. A YouTube interview with METR researcher Ajeya Cotra provides an even more direct sense of how serious the issue has become. What exactly needs to happen to make skeptics realize that those working in the AI industry—and indeed all of us—truly have reasons to be uneasy?

Here's a question: Who reviews a company's quarterly report before it's submitted to the U.S. Securities and Exchange Commission (SEC)? Do senior executives, especially core leaders holding substantial equity stakes, pay attention to significant new information concerning themselves in these filings? Oracle disclosed its latest quarterly filing with the SEC last Friday, and a peculiar incident surfaced, raising precisely this question. The filing revealed major news: Executive Chairman Larry Ellison established a stock trading plan in June to sell up to 50 million shares of his 1.158 billion Oracle shares by October 24. Upon completion of this sale, his ownership stake would have slightly decreased from approximately 38% to 36.7%.

However, the very next day after the disclosure (Saturday), Oracle issued a statement: Ellison had cancelled the trading plan, had not sold any shares, and had no new intentions to reduce his holdings. What exactly happened? Did Ellison change his mind about selling without informing the subordinates responsible for preparing the SEC filing documents? This scenario isn't entirely implausible, especially considering that Oracle's stock price experienced a significant drop after the trading plan was established, though it has since recovered most of its losses. Regardless of the reason, the review process for future filing documents must be more rigorous going forward.