Wall Street is under minor AI-Nxiety attacks. Long-standing concerns about the large, perhaps undisciplined high-tech enterprise investments in artificial intelligence infrastructure and the gorgeous evaluations awarded to these businesses began to boil last week. And with that, there is a fierce debate over whether a dangerous fairness bubble is ongoing. The Wall Street Journal provided an in-depth front-page treatment with AI Capex Binge, drawing a clear but related comparison with the broadband internet epidemic of the late 1990s, narrowing his eyes skeptical of the possibility of retrieval. Michael Cembalest, a respected and large picture strategist at JP Morgan Asset Management, has published a report described in the Vintage-Horror-Film Style, known as the “Data Center Blob.” The opening line is intended to set scary tones. AI-related stocks account for 75% of the S&P 500 return. David Einhorn, hedge fund manager at value-oriented Greenlight Capital, flagged the prospect of “massive capital losses,” with multiple articles noting the increased use of debt funds for building and server rack payments (Oracle sold quickly in paper last week, and tech companies issued $15.7 million in debt this year). GQG Partners is a $172 billion investment manager founded in 2016 by veteran investor Rajiv Jain, pursues a “quality growth” strategy and published an investor letter called “Dotcom for Steroids” two weeks ago. In recent years, the company owns Nvidia and other high-tech stocks, declared, “For the first time in our company’s history, many of today’s large tech companies, especially those with meaningful roles in AI infrastructure buildouts, represent quality that appears to be backwards. The paper shows that digital advertising and cloud services are growing as the key markets driving large-scale technology revenues mature, competitive and growing, leading to digital advertising and cloud services growing. (Side note: Few people realize that 2026 return rates for alphabet and meta platforms are projected to slow down to the 6-8% range.) GQG pushes back that today’s high-tech sector is higher quality and not overvalued. With GQG’s approval, we can see that this careful call can be very early, and from here it means quite upside down. Wall Street strategists generally resist bubble chatter, and Deutsche Bank has stated that Hyperschool has enough cash flow to invest well while backing up inventory, and Barclays is far below the leveraged telecom player player of textile optics’ frenzy 25-30 years ago. The discussion won’t resolve here now, or perhaps anytime soon. The bubbles are rare, and are recalled and misidentified much more frequently than they occur in the typical bull market. You may not see any mania comparable to the 1999-2000 crescendo, so there is a general pastime in which we map the current rise to the late 90s and now, for a few years, a large distance from that peak could become a trap rather than a guide. Meanwhile, this latest crisis of faith on the subject of AI is quite likely to be something that is necessary to settle down a bit with investors, rebuild the slight barrier of worry, stress-testing growth models and then extending fun. When should you really start worrying? Which opened up the question of how to know when to start worrying about hilarious extremes? The initial state is truly extraordinary market performance that feeds itself by climbing at a steeper angle than we saw in the current phase. The benefits of pure AI names and the wild speculative things around the edge (evolution of quantum computing, cryptocurrency shell companies, nuclear power generation plays) were extraordinary, but the wide index rating is good, but not extreme. The current bull market acquisition, where 80% of the S&P 500 is three years ago, is in the middle of the pack. The current total annual revenue for the next five years is 16%. The end of 1997 was 20%, and the end of 1998 was 24%. The starting point was that the market was more expensive and highly owned, as this cycle was both high in valuation and wide public ownership of stocks. This should mean muted equity returns in the next decade, not time-tested, not iron-covered market mathematics. But it doesn’t need or imply a poised bubble that bursts. Ned Davis Research last week said its number on the long-term relationship between household stock ownership, with some of the financial assets (now record-high) and subsequent S&P Running as a 500 performance. On this chart, the annual returns for 10 years are inverse scale. The relationship has been close for decades, but over the past decade it has slacked in favor of better returns than history suggests. Another feature of the market growing towards a volatility is its rise in volatility along with stock prices. The volatility trend of excitability in 1999 promoted such a phenomenon, with CBOE’s volatility index not falling 20% by 20% and NASDAQ composites rising 85%, making it rarely below 20. Today, VIX has only been above 20 days with stable index gains for the past four months. This is not that the current market lacks some volatility abnormalities, but that they are more subtle. The correlation between S&P 500 stocks over the past three months is 10 years less and suppresses index movement, but surprises things under the surface. The bespoke investment group is currently the only group of stocks that are currently on a strong upward trend, and is the S&P It is the 10% most volatile of the 500, and Goldman Sachs Equity Desk said it pointed out on Friday. 2017 (measured on a 3M realisation basis). There was no particular concern right after April 2017. The market remained in a close uptrend for another nine months. In February 2018, pent-up energy burst with a “Volmageddon” correction. A general distinction drawn between the 2025 environment and the favourite bubble moments of all people’s favourite bubble moments in the era of Pets.com is the slow pace of IPOs and other stock issuances. According to Dealogic, total stock supply this year was $180 billion. It is true that the sketchy IPO rush ultimately overwhelmed the tapes in 2000. Still, today’s capital markets are becoming very messy, with these penny stocks hijacked by warehouse cryptocurrencies, making up a large portion of investor capital, chasing borrowers in search of a comeback of 2021-vintage meme stocks. This is more of a territory of tactical concern at this point, rather than evidence of some kind of generational peak formation. And bubbles – expansion or contraction – are not necessary to make the market wobble in bursts or turbulent flows. Other notable features of the short-term setup still include addiction tapes. The S&P 500 is even rejected by 3% pullback in almost five months. .spx 6m Mountain S&P 500, six months, the S&P hit a record high of 6,699 twice last week, suggesting that 6,700 people are considered to be the right selling level. Core AI beneficiaries – Oracle, Micron and Broadcom have poor news flows despite star results and positive news flows, and Nvidia is stuck at the late July level. The wide picture of the Fed leaning towards a lower rate in a stable economy is uninterrupted, but some non-committed Fed talk and sideways inflation have retarded market certainty for 10-year Treasury yields back to three-week highs, and two more interest rate cuts this year. Despite the market refused to assert calendar-based weaknesses in August and September, it could theoretically recite the seasonal softness that opens the windows of pullbacks to October. One thing to note in the coming months is Street Strategists, which updates their 2025 index targets and deploys 2026 phones. Currently, the median bogeys for 2025 are correct for current levels, implying a beneficial reservoir of attention. However, at the end of last year, handicappers came out passionately at their 2025 target. This sent high expectations even before the April tariff shock crushed the tape and finally set the bear traps, setting investors on a nasty early setback.
