Jim Cramer rarely minces words. On a recent episode of CNBC’s “Mad Money,” he delivered one. “You are unsafe at any level.”
The veteran market commentator pushed back hard against growing chatter that 2026 echoes the final months of the dot-com frenzy. He sees something different. Something worse. Investors chase a handful of artificial intelligence winners with abandon. They dump everything else with a speed and ferocity he says exceeds anything from 1999.
The Narrow Trade That Defines 2026
Cramer spoke on May 11 as the S&P 500 and Nasdaq posted fresh record closes. Enthusiasm for semiconductors, data centers and all things AI powered the gains. Yet the breadth told another story. “We keep hearing this drumbeat that 2026 is 1999 all over again,” he said on the show. “But the difference between now and 1999 is that this market does not stop punishing the companies that disappointed … You are unsafe at any level.” (CNBC)
The numbers back his point. Semiconductor stocks alone accounted for roughly 70% of the S&P 500’s $5.1 trillion increase in market value this year, according to Reuters data cited in coverage. The Philadelphia Semiconductor Index jumped 64% since late March. The so-called Magnificent Seven — Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta and Tesla — now represent about one-third of the entire S&P 500’s market value. (Yahoo Finance)
And the punishment for those left out? Swift. Abbott Laboratories, a stalwart of American health care, sank 34% year-to-date after a narrow earnings miss. Danaher fell 27% amid what Cramer called “a savage string of not-so-great quarters.” Boston Scientific, Intuitive Surgical, Medtronic, ResMed, Stryker and Zimmer Biomet all hit new lows. “This is Abbott Labs for heaven’s sake,” Cramer said. “A market that punishes Abbott Labs is a market that despises anything not connected to tech and the data center.”
Portfolio managers, he observed, appear to have drawn a bright line. Stocks tied to AI and data centers enjoy voracious demand with seemingly little sensitivity to economic cycles. Everything else faces abandonment. The selling carries “a level of fear I can’t ever remember seeing before.”
But Cramer doesn’t stop at description. He argues the current setup creates hated stocks that are over-hated and loved ones that are over-loved. The loved group clings to the data center narrative. The hated group includes established names in health care and software seen as vulnerable to AI disruption. Investors, he says, have grown quicker to jettison any company that fails to tie itself convincingly to the AI boom.
Recent coverage reinforces the concentration risk. A Reuters report highlighted how software and health care names in particular have suffered from fears of AI-driven obsolescence. Meanwhile, IPO enthusiasm continues. Cramer recently cautioned against chasing Cerebras Systems after its blockbuster debut pushed the valuation to $100 billion, warning that such over-enthusiasm “ends badly.” (CNBC)
Even within the favored group, selectivity matters. Cramer has flagged individual AI-related names as “too high” at various points this year, including Vertiv Holdings in April and others showing parabolic moves. Yet he maintains that the broader data center and cloud infrastructure theme retains structural support. Nokia, for instance, earned a buy call from him on the same May 11 show because of its dual exposure to cloud infrastructure and defense contracts.
The comparison to 1999 invites scrutiny. Back then, the Nasdaq’s surge rested on internet infrastructure plays and speculative dot-com names. Many traditional growth stocks outside tech suffered, but the punishment often came in waves rather than instant, unrelenting drops. Today’s market, Cramer contends, shows less mercy. Disappoint once and the stock may never recover its prior multiple. The fear level feels higher. The conviction around the winners feels more absolute.
Market watchers point to additional pressures. Geopolitical tensions, including the ongoing situation with Iran, have raised concerns about economic outcomes. Jamie Dimon of JPMorgan Chase recently noted that such conflicts increase the odds of adverse results, a theme echoed in Cramer’s own commentary on consumer and retail names. Stocks like Chewy have been hurt by war-related narratives even when their fundamentals hold up.
Still, not every voice agrees the end is near. Some strategists argue the AI boom retains room to run. Dan Niles of Niles Investment Management suggested in a recent CNBC appearance that a market bubble “can inflate a lot more before the end.” Such views clash with Cramer’s call for caution.
Investors face a bifurcated reality. Broad indexes hit records. Large portions of the market languish. Concentration at these levels has historically preceded sharp corrections, yet the drivers — genuine advances in AI capability and massive capital spending on data centers — differ from the vaporware of 1999. Cisco today, Cramer noted in related remarks, bears little resemblance to its dot-com peak self.
So what should professionals do? Cramer’s message isn’t outright bearish on the indexes. He simply warns that owning the wrong names at the wrong time carries extreme downside. The loved stocks may need to “catch their breath.” The hated ones could be oversold but require patience and proof that the AI disruption thesis doesn’t apply.
Portfolio construction matters more than ever. Diversification outside the narrow AI trade has been painful. Yet blind chasing of the winners at current valuations invites its own risks. “The hated are over-hated and the loved are over-loved,” Cramer concluded. That tension defines this market. And it shows no sign of easing soon.
Additional recent reporting from Benzinga and GuruFocus has echoed Cramer’s concerns about overheated tech and the need for balance, citing healthcare as an area hit particularly hard by sector rotation. (Benzinga) The data keeps coming. The debate intensifies. For industry professionals, the message is clear. Respect the narrowness. Prepare for volatility. And remember: in this environment, safety is never guaranteed.
WebProNews is an iEntry Publication