This latest iteration of the AI hype train features massive rallies in stocks like Oracle and Broadcom. Both surged after reporting quarterly results that specifically spelled out how artificial intelligence will drive sales and profit in the coming quarters.
On the surface, these massive gains look like one more sign that the AI bubble is ballooning larger. While there could be some credence to that argument—particularly considering that an OpenAI investment helped boost Oracle—the market’s distinction between winners and losers during second-quarter earnings season has taken on a new dynamic. Specifically, the AI trade is now officially a “show me” story. The winners are companies that can demonstrate sound fundamentals in their AI business.
Strategists at Goldman Sachs recently highlighted this trend in a note to clients. When analyzing a group of stocks they call “AI enabled revenue” companies—mostly software stocks that could see revenue boosted from increased AI adoption—the strategists found a wide dispersion of returns during second quarter earnings season.
For example, popular AI stock plays MongoDB, Digital Ocean Holdings, Snowflake and Commvault Systems all rallied at least 18% the day after reporting earnings. Meanwhile, Spotify, Fortinet, Confluent and the Trade Desk all fell 12% or more in the trading session following their latest quarterly releases. A key determining factor for the divergence proved to be forward guidance of their AI businesses.
“What’s remarkable about the whole AI trade, from semiconductors to data centers to software all the way to the rest of the U.S. corporate space, is how disciplined the market has been in trading concrete near term earnings,” Goldman Sachs analyst Ben Snider told Barron’s. “We talk all the time with investors about the long term productivity implications of AI and what it will mean for the U.S. economy in the next decade or two, and what it will mean for corporate earnings in the next decade or two. But investors are really trading the earnings impact of AI today.”
This theory has been on clear display during recent market action. Companies that aren’t seeing AI drive a large inflection in growth are paying the price. Take Salesforce for example. The Dow component and one of the 50 largest stocks in the S&P 500 has long been identified as an AI winner as investors embrace so-called AI “agents.”
In 2025 that has been far from the case. Salesforce shares are down more than almost 30% this year. After a roughly 30% rally in 2024 as Wall Street cheered the software company’s AI story, Macquarie’s head of U.S. Software & Services Research Steve Koenig told Barron’s Salesforce investors likely “got ahead of themselves,” when thinking about how the emerging technology could boost near-term results.
“Salesforce has suffered from overly high expectations about what AI was going to do for them,” Koenig said, noting that Salesforce itself doesn’t believe AI will significantly boost revenue until at least next year.
Adobe, also among the 100 largest companies in the S&P 500, has seen similar struggles. Despite clear paths for AI agency use cases in Adobe’s editing applications, the stock is down over 20% in 2025 and more than 30% in the past year. Adobe shares even dropped by almost 2% over the last month, during a period when the S&P 500 has risen over 2.4% and the Information Technology sector has added about 2.7%.
Jefferies internet and software analyst Brent Thill told Barron’s the problem for stocks like Adobe and Salesforce is there hasn’t been a sharp move higher in fundamental metrics. Unlike Oracle, which recently told investors it expects cloud infrastructure revenue to grow from $18 billion this year to $144 billion in its 2029 fiscal year, Adobe and Salesforce aren’t projecting explosive growth.
“There’s no S Curve,” Thill said. “There’s no inflection in the numbers.”
For investors to turn to this next leg of the AI trade there’s need to be a “much bigger inflection before people are going to get excited.”
The AI trade still feels like it’s everywhere and is a key reason that firms like Evercore, RBC, Morgan Stanley and Wells Fargo have called for the S&P 500 to rally above 7,000 by the end of 2026. But investors remain selective. Even popular retail names of the past few years, such as C3.ai or SoundHound, find themselves out of favor. C3.ai has lost nearly half its value this year. SoundHound is down almost 30% in 2025.
While there will undoubtedly be parts of the AI cycle that prove to be overdone, current market action indicates it might not be time to call a bursting of a bubble just yet. Instead, the market seems to be deciding AI winners and losers like it always does. It’s all about the expected pace of future profit growth.
“There are a lot of things we can look at in the market today that are reminiscent, in some ways, of past bubbles,” Snider said. “But when I look at valuations, and I look at the implied long term earnings growth rates, those do not look at all like a bubble today.”