Better Stock-Split Artificial Intelligence (AI) Stock: Nvidia, Broadcom, or Lam Research?

With the artificial intelligence revolution in full swing, share prices of favored semiconductor stocks have been on the rise. In late May, I identified five AI stocks that were due for a stock split.

Lo and behold, three have already announced splits since then: Nvidia (NVDA -1.91%), Broadcom (AVGO -0.34%), and Lam Research (LRCX 0.01%), with Nvidia having already executed its 10-for-1 split on June 7. Broadcom will split its stock 10-for-1 on July 15, and Lam will execute its split on Oct. 3.

While retail investors and employees may more easily be able to buy shares of a lower-priced stock, a stock split doesn't change fundamentals or overall business value. It's just like cutting a pizza into smaller slices: You still have the same amount of pizza!

But after their post-split surges, which AI winner is the better buy now?

How Nvidia, Broadcom, and Lam Research benefit from AI

Nvidia has been the obvious winner from the AI revolution, with market-leading GPUs and CUDA software which allows GPUs to be programmed for data processing. As a first-mover, Nvidia has had a virtual monopoly on chips needed for training large language models thus far.

Broadcom is an ethernet networking chip leader enabling the systems that shoot massive amounts of data between memory and processors in and around the data centers. Furthermore, its custom ASIC (application-specific integrated chip) IP business helps large cloud companies make their own in-house AI accelerators. Needless to say, both of those businesses are booming.

And Lam Research is a key semiconductor equipment manufacturer which dominates certain etch and deposition process steps, enabling the production of Nvidia and Broadcom chips, as well as many more.

Valuation

All three stocks of have surged over the past 18 months, as the semiconductor downturn of 2022 gave way to the AI enthusiasm of 2023. While none of their stocks are "cheap" at the moment, but there are some notable differences in valuation.

Data source: Yahoo! Finance

Nvidia has the highest forward P/E ratio, but also the highest expected growth rate and the lowest PEG ratio, as defined by forward P/E divided by the expected earnings growth rate over the next five years.

So, one could actually make the argument that despite Nvidia's amazing recent run, it's still the cheapest of the three AI stock split stocks.

But investors may want to be cautious

Of course, one should remember that a PEG ratio is based on the expected growth rate over a number of years, and thus subject to a lot of uncertainty. After all, before the introduction of ChatGPT and the ensuing AI mania, how many would have anticipated a near-50% growth rate for Nvidia for five years?

Meanwhile, an expected growth rate doesn't factor in how risky or uncertain that forecast may be. Should anything go wrong with either the current AI spending boom or Nvidia's competitive moat, its growth rate in a few years may not match these lofty expectations.

In terms of AI spending, we are by most accounts still in the early innings of the revolution. But if companies don't find they are getting a return on all of that spending, AI investment could slow. While many organizations are reporting strong benefits from AI, not all AI projects are working out. For example, McDonald's just folded its AI drive thru experiment and is rehiring humans, at least for now.

Image source: Getty Images.

Moreover, Nvidia's net margins surged to 58.5% last quarter, an outrageous number for a chip company. With profits that high, one can be sure Nvidia will see intense competition for AI chips going forward, not only from its main merchant semiconductor competitors, but every cloud company that is also now building their own custom accelerators to save costs.

If AI investment slows down or significant competition begins to eat into Nvidia's growth or margins, its projected earnings growth might not live up to the current five-year projection.

Broadcom and Lam Research could surprise to the upside

While Nvidia is seeing a lot of AI-powered growth right now, Broadcom and Lam Research should see a wave of growth going forward. Networking investment usually comes after chips are bought, benefiting Broadcom, and as chipmakers begin to invest in new capacity, they will buy more production machines, benefiting Lam.

Meanwhile, Broadcom and Lam Research have more diverse businesses and greater services segments, mitigating some risk.

Interestingly, those non-AI businesses are coming out of a downturn right now. For Broadcom, its broadband and storage controller businesses reported 39% and 27% declines, respectively, last quarter. But management said it believes these segments have bottomed. Meanwhile, Lam Research's largest business used to be NAND flash storage, which has seen its biggest downturn ever and hasn't yet recovered.

Lam has done an excellent job of gaining market share in foundry and logic chip production to counteract the NAND flash downturn, and Broadcom has pivoted away from cyclical chip markets in general to focus on software, with its recent massive VMware acquisition looking like a home run.

Both Broadcom and Lam also have big chunks of their business that are less volatile than chipmaking. Broadcom's software division now makes up 42% of its revenue following the VMware acquisition. Meanwhile, Lam's spares and services business, a less-volatile business largely tied to its installed base, amounted to 37% of revenue last quarter. While Nvidia is also developing software offerings today, that only amounts to low-single-digit percentage of its sales, as it still gets the vast majority of its revenue from AI chip sales.

After a big run, play it safe

Broadcom and Lam Research both operate in oligopolies, with Broadcom really only having one main competitor in accelerator ASICs and networking chips, while Lam tends to compete with just one or two etch and deposition competitors, depending on the manufacturing step.

Nvidia appears to be a monopoly today, but will see the most competitive threats going forward, not only from rival merchant chip companies, but also the cloud companies making their own AI accelerators in-house.

Thus, the preference today should be for the relative safety of either Broadcom or Lam. Both companies have more diverse businesses, a higher percentage of earnings coming from software/ services, and cheaper valuations with lower expectations. Moreover, Broadcom and Lam should benefit from AI growth no matter which chipmaker dominates the industry's future, whereas Nvidia will have to defend against myriad competitors coming for it over the next five years.