Broadcom’s blockbuster earnings report has rekindled a long-running debate among investors: Just how firm is Nvidia’s grasp on the AI chip market? The problem, though, is that might not be the best question for investors to ask. Instead, investors are better served asking just how large the AI computing market itself can grow to be. As investors in Nvidia and Broadcom, we clearly see a role for both chipmakers to play in the AI race. Nvidia is the dominant provider of “merchant” silicon, selling the same advanced chips to various data-center operators, along with a rich software presence. Broadcom is the leading purveyor of “custom” chip-design services, utilized right now by a small subset of deep-pocketed tech companies like Google that, generally speaking, are trying to get more out of their own software stacks. Undoubtedly, some sales that could’ve gone to Nvidia will flow to Broadcom’s coffers in the years ahead as internet giants like Meta Platforms , a fellow Club name widely assumed to be one of Broadcom’s existing custom-chip clients, look to run certain internal workloads at lower costs using specialized chips. Some cloud-computing providers may also look to diversify the kind of computing power they can offer customers. The tension is that, if this dynamic indeed plays out, will it mean that Nvidia is no longer be able to meet Nvidia investors’ growth expectations? Or, is the demand for AI infrastructure large enough over the coming decade that Nvidia can keep growing sales and earnings, even as its own market share comes down? Our bet is on the second scenario. A pair of notes out Monday morning go to the heart of this debate — one comes from the analysts at Citigroup, the other from Melius Research. Citi cut its 2026 sales estimate for Nvidia by $12 billion and, in turn, its price target on the stock, directly citing Broadcom’s comments last week about rising demand for its custom AI chip services. The firm went to $200 a share from $210, a move certainly counter to what we have seen in recent years when it comes to Nvidia price targets, which have tended only to go up. On the other hand, Melius reiterated its price targets on both stocks and argued that even with Broadcom seeing an incredible amount of interest in its custom silicon solutions, the overall demand is large enough that both companies stand to grow immensely in coming years and likely outpace investor expectations. To be sure, while the Citi analysts cut their price target, they are still positive on the name. They maintained their buy rating and noted that even their reduced 2026 estimates are above the Wall Street consensus. Nevertheless, a bullish firm getting somewhat less bullish is notable and it underscores the current debate on the Street. Do you reduce your Nvidia exposure to increase your Broadcom stake? We think the real answer is simply to own both, planting our flag in the Melius camp. While Broadcom is certainly cementing itself as a top AI stock to own, our belief is the demand is simply so great that both will win. Do the homework on both. Adjust their individual weightings based on your standard portfolio management disciplines — i.e., if one has gone on a parabolic move, consider booking some profits; the same goes for it a stock exceeds your threshold for weighting, which for the Club is about 5% on any given position. But analyzing the situation by only playing one against the other is short-sighted. What it really comes down to is the size of the total addressable market, often abbreviated to TAM. Sometimes, there will be a company that has so much market share of a new, rapidly growing industry that it is a given they will lose share as others see how much the company is making and look to compete. However, as the pie gets bigger, they can lose some of that share and still continue to grow. In other words, you get a smaller share or percentage of a larger pie, but that smaller share ends up being larger than the entirety of the initial, smaller pie. NVDA AVGO 1Y mountain Nvidia and Broadcom’s stock performance over the past 12 months. Consider the case of Amazon and the online shopping market. Back in June 2018, the research firm eMarketer said Amazon ended 2017 with about 44% of the U.S. e-commerce market, but was on its way to capturing just under 50% in the following year. As of 2023, eMarketer pegged Amazon’s share of US ecommerce at about 40%. Nonetheless, Amazon’s sales in North America went from $141.4 billion in 2018 to $352.8 billion in 2023. They’re on track to be $424 billion this year, according to estimates compiled by FactSet. The point is that while market share is important and the size of the market is an important metric to consider when seeking to determine the size of the opportunity, you do not want to get caught up thinking that you are strictly investing in market-share growth. Instead, what you’re investing in is growth in the company’s sales and earnings. Since the end of 2017, Amazon has returned around 304% versus 175% for the S & P 500, including reinvested dividends. When we apply this to Nvidia and Broadcom, we think a similar argument makes sense. Nvidia, as a first mover in the AI semiconductor space, has enjoyed the benefits of being the dominant player in the market. However, as more companies started to see the potential magnitude of AI demand, it is only natural that they would move into the space — which, in this case, means looking to companies like Broadcom to help them design their own AI chips. We’ve also seen AMD ramp up its efforts to compete with Nvidia on the “merchant” silicon side of AI. This doesn’t mean Nvidia will stop growing. After all, the company can barely keep up with the demand it has now, and all signs point to the need for AI infrastructure only increasing from here. “We think there are signs that the AI compute/networking TAM is entering a hugeness that is hard to fathom,” Melius Research wrote in its Monday note. By their estimation, the serviceable addressable market (SAM) stands at about $2 trillion toward the end of the decade — that’s about half of the total $3 trillion to $4 trillion in data center capital expenditures that Nvidia CEO Jensen Huang recently predicted would occur by 2030. Against that backdrop, Melius argued that both Nvidia and Broadcom “are much more likely to beat our 2027 estimates than not. In fact, if Broadcom can get just 20% of our $2T SAM estimate for 2030 and Nvidia keeps just 40% then both stocks are going a heck of a lot higher.” And that’s really all that matters at the end of the day – the totality of AI demand. Indeed, Stacy Rasgon of Bernstein Research made a similar case Friday when he appeared on CNBC’s “Closing Bell.” At that time, Broadcom shares were soaring in response to its earnings report the prior night, and Nvidia was down a few percentage points. “I don’t think the right question right now is necessarily who is winning or losing,” Rasgon said, while noting this isn’t the first time Nvidia shareholders fretted custom competition. “I actually think, personally, the question is better off to be put: Is the opportunity in front of us still large, or is it not? I think [custom chips] will take share. They’re coming from a smaller base. I don’t think they dominate. And I think if the opportunity in front of us is still big, if we’re still early in this, as I think we are, I think they can both thrive. Think about it this way: If the opportunity in front of us is not still big, like, they’re both screwed. That’s the right question right now, I think, not as much who is winning or losing.” Bottom line While the market makes a lot of noise about which AI chip stock to be in based on which CEO spoke most recently — or reminded the world that AI demand has plenty of room to run — we think investors will be better served keeping their eye on just how large the entire market will grow to be in coming years and targeting the names with best-in-class offerings. In the case of merchant AI chips, that’s Nvidia thanks in large part to a massive software ecosystem that serves as a competitive moat against other competitors like AMD. And in the case of custom silicon, it is Broadcom. (Jim Cramer’s Charitable Trust is long NVDA, AVGO, META and AMZN. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
