
Many investors appear bullish on stocks and the artificial intelligence trade early on Thursday, with Anthropic ‘s Nasdaq debut expected to break Wall Street records at market open. But for one seasoned investor, Nov. 19 is the day to cash out of the equity market entirely. “This is it,” Joachim Klement, a research analyst at Panmure Liberum, said in a note on Thursday morning. “Why I am selling.” Klement confirmed to CNBC that he would be selling all stocks in his personal holdings, not just those listed in the United States. “The S & P 500 has broken below its 200-day moving average, and we are moving into a fully defensive position,” his team wrote in Thursday’s note. “While this indicator is not foolproof, market dynamics have turned decisively bearish, in our view.” Klement, a former chief investment officer who has worked in financial markets for two decades, has been warning of an AI bubble since May. He said he had become even more bearish since the end of the summer. “I have always emphasized that my personal trigger is the S & P 500 relative to the 200-day moving average,” he reiterated in Thursday’s note. “If the S & P 500 drops below the 200D MA, I will sell equities and move into long-term government bonds. If it then moves above the 200D MA again, I will move back into stocks until the S & P 500 drops below the 200D MA.” Klement conceded that this strategy “may incur a lot of portfolio churn,” but said he would be “out of stocks when the AI bubble bursts and back in if setbacks are temporary in nature and not persistent.” “At the end of European trading, I will sell my personal holdings in stocks and move into long-term government bonds,” he said. “I may be wrong, and this is not the bubble bursting, but if that is the case, my trading rule will ensure I go back in without too much opportunity cost.” Bond markets have also been volatile this year, with debt issued by governments across the globe selling off aggressively as the Iran war fed expectations of inflationary pressure and more hawkish central banks. Klement’s team at Panmure Liberum first warned about the risk of an AI bubble bursting back in May, after they examined the AI capex boom. Panmure Liberum’s analysis suggested a reversal of the boom could lead to a drop in stock markets of 30% or more. In September, the group’s analysts predicted the AI bubble would burst in either 2027 or 2028, pricing the chance of that happening next year at almost 50%. A month later, they increased the likelihood of an AI crash in 2027 and introduced a price target for the S & P 500 of 5,000 points for the end of 2027. That represents a retreat of 35% from where the index was trading at the time. On Wednesday, the S & P 500 closed at 7,323.63 points. .SPX YTD line S & P 500 “Today, we recommend investors go fully defensive as price momentum has, in our view, decisively shifted to the downside,” Panmure Liberum’s analysts wrote on Thursday. Panmure Liberum has been observing two other triggers to assess whether the AI bubble had run its course: capex by AI hyperscalers and Anthropic’s IPO. “In recent weeks we have seen Microsoft aggressively cull its deals in neoclouds and focus its capex more on self-built data centres,” Klement wrote. “This is not yet a reduction in capex vs 2027 expectations, but we still think that capex guidance for 2027 could be lower than current expectations. And every dollar less in capex vs expectations means one dollar less in revenues for the semiconductor supply chain.” His team added in the note that it’s “clear” from Anthropic’s regulatory filings that “revenue growth is slowing down both relative to compute costs and, more importantly, in absolute terms.” Anthropic priced its IPO at $83 per share on Wednesday, and is selling about 980 million shares, bringing the amount raised in its stock market debut to a record $81.3 billion. “We do not know how widespread Al risk is, given that Al debt is not just the bonds of hyperscalers and other major companies,” Klement also warned on Thursday. “A lot of the Al debt has been packaged as private credit, where there is no transparency about how much of it is around and who the ultimate investors are.” AI hedges For investors considering a pivot away from AI stocks but wanting to remain in the equity market, Klement said his team’s analysis shows “U.K. stocks are the best hiding place from a regional perspective.” They analyzed the 10 worst days for the Goldman Sachs Broad AI Basket of stocks this year, and looked for European companies with the highest average returns around the days when the AI stocks sold off the most. Klement’s top 10 companies, or “Best AI hedges,” in London’s FTSE 100 were: GSK , AstraZeneca , Unilever , Experian , Diageo , Sage , Haleon , Relx , Reckitt Benckiser and Croda . Other companies that performed well in the analysis more broadly across Europe were Pernod Ricard , L’Oreal and Beiersdorf . “We use the S & P 500 vs its 200-day moving average as the signal to assess if price momentum has changed,” Klement said. “If the S & P 500 drops below the 200D MA, we take this as an indication that the market is telling us it is exhausted. Every investor knows that this signal is not perfect and it may well be a false trigger.” He noted that the S & P 500 crossed the 200-day moving average more than ten times in each of the last bubble tops before the bear market accelerated. “Today is thus not the beginning of the end, but the end of the beginning,” Klement wrote. “In the near term, we expect heightened uncertainty with little gain in either direction and a market getting increasingly exhausted.”