Jim Cramer AI Chip Investment Advice: Why He’s Telling Everyone to Buy the Dip on Broadcom

It was mid-September in San Francisco, and shares of all AI-related stocks were plummeting, creating an almost funereal atmosphere. The CEO of Anthropic recently advocated for a slowdown in the development of frontier AI. Regarding spending, OpenAI was giving conflicting signals. Jim Cramer, who was broadcasting from the West Coast, entered that mess, telling his audience that what they were witnessing was a clearance sale rather than a collapse while staring at red screens. Depending on what transpires next, that kind of confidence could appear brilliant or reckless, and it’s worthwhile to look at the argument he’s making for investment in AI chips at this time.

More than any other single name, Broadcom has been Cramer’s focus. AI semiconductor revenue increased 221% to $16.7 billion in the chipmaker’s recently released fiscal third-quarter results, which showed 86% revenue growth year over year. For a $1.6 trillion company, those figures are abnormal. These are the kinds of numbers that ought to cause a stock to rise rather than fall.

However, the market penalized Broadcom when its fourth-quarter guidance came in at $34.8 billion, which was marginally less than analyst expectations of $35.03 billion. Over the next few weeks, the stock fell by about 10%. On the September 14 episode of Mad Money, Cramer made a simple point: investors were mistaking a small guidance error for a failing company, and the two had nothing to do with one another.

What transpired that same evening gave Cramer’s pitch greater weight than a standard contrarian call. The AI slowdown narrative was directly addressed by Broadcom CEO Hock Tan during his appearance on the show. Tan didn’t hedge when Cramer questioned whether infrastructure spending would be slowed by safety concerns from businesses like Anthropic. The demand for compute infrastructure, according to him, is “extremely strong and durable.” Tan also reiterated his forecasts, which call for yearly sales of AI chips to reach $115 billion by fiscal 2027 and $230 billion by 2028. These ambitious goals are supported by a $179.2 billion backlog of unfulfilled performance commitments. The discrepancy between the CEO’s public statements and the stock’s market price is difficult to ignore.

Although Broadcom is undoubtedly the focal point of Cramer’s larger AI chip investment thesis, it goes beyond that. Instead of focusing on the companies creating the models themselves, he has been directing investors toward what he refers to as the “infrastructure layer”—the businesses creating the actual hardware that AI operates on. With a forward earnings multiple that he believes is undervalued considering its dominance, Nvidia continues to be his top choice. Due to the unquenchable demand for high-bandwidth memory in data centers, Micron has also appeared frequently. His commentary has also included Applied Materials and ARM, indicating a conviction that sustained AI buildout benefits not only the top chip designers but the entire semiconductor supply chain.

This story has a central tension that needs to be honestly acknowledged. Approximately 71% of Broadcom’s anticipated fiscal 2027 and 2028 custom chip deployments may rely on OpenAI and Anthropic, two businesses whose executives have recently discussed stepping back or moving more cautiously. Broadcom’s ambitious revenue targets become much more difficult to meet if those companies slow down their orders. In order to mitigate that risk, Cramer essentially relies on Tan’s order book rather than public declarations regarding AI ethics. Perhaps that’s right. However, investors should acknowledge that it is a wager rather than a guaranty.

For the time being at least, the institutional money appears to be stable. 170 hedge funds held Broadcom positions, down just three from the previous quarter, according to filings monitored through the second quarter. With more than 15 million shares, Fisher Asset Management continued to be the biggest owner. Following the earnings report, Morgan Stanley increased its price target to $505. The average analyst target is close to $518, indicating significant upside if the company meets even a small portion of its multi-year guidance. Wall Street as a whole hasn’t given up on the stock; it’s simply awaiting the next set of data to support the theory.

As you watch Cramer solve this in real time, you’ll notice a recurring pattern. When the market is in a panic and he advises people to think instead of act, he is at his most captivating. Whether or not businesses continue to invest in AI infrastructure at the rate that everyone is anticipating will determine whether or not his investment recommendations regarding AI chips turn out to be accurate. This is something that no television personality can control. They will, according to Hock Tan. They will, according to the order book. However, the market isn’t entirely persuaded just yet. The opportunity either lives or dies in that space between confidence and skepticism, and it’s precisely the kind of