Investing

3 Winners From the Next Massive AI Spending Boom (Hint: It's Not NVIDIA)

Eric Bleeker

Published: July 5, 2024 8:10 am

Another week, and more good news for AI investors. Last week it was revealed that memory maker SK Hynix plans to invest $74.6 billion into expanding its capacity to make chips for the AI boom. While rabid demand for High-Bandwidth Memory (HBM) gets less attention than stocks like NVIDIA (Nasdaq: NVDA), it is nonetheless one of the most compelling growth areas in artificial intelligence investing. We dive into winners from this red-hot AI market. Companies discussed include SK Hynix, Micron (NASDAQ: MU), ASML (NASDAQ: ASML), Taiwan Semiconductor (NYSE: TSM), and BE Semi.

Winners from the Memory Spending Boom

Below you’ll find some of the highlights from the discussion between 24/7 Wall Street Analysts Eric Bleeker and Austin Smith:

• Today we’re focusing on the recent news that SK Hynix plans to invest $74.6 billion in semiconductor across the next 5 years.

SK Hynix, the leader in AI memory chips called HBM (high bandwidth memory), plans to invest 103 trillion won (US$74.6 billion) in capex over the next 5-years (thru-2028) on new semiconductor factories and production line equipment, media report. 1/2 $MU #Samsung $ASML $AMAT $LRCX…

— Dan Nystedt (@dnystedt) July 1, 2024

• SK Hynix’s Chairman had this to say about the dynamics leading to this aggressive spending plan: “The winds of AI-related change are so strong in the U.S. that there is nothing to talk about but AI.”
• High-Bandwidth Memory is a booming market thanks to AI. It’s now widely understood that GPUs (the type of processor made by NVIDIA) are uniquely suited to training AI models. What’s less understood is the amount of memory required for AI training.
• Neural network training requires tons of memory. You have to store input data and parameters as inputs work through a network’s layers.
• The unique memory usage qualities of AI have led to booming demand for High-Bandwidth Memory, which is the current memory technology best suited for AI servers.
• High-Bandwidth Memory is just one of many industries that are seeing a boom after NVIDIA.
• That is to say, while NVIDIA saw massive growth as investors first scooped up GPUs (the processing units), now other aspects of the data center are booming. That’s been reflected in rising sales for areas like HBM, liquid cooling, and networking. There are even more areas that are still in the first innings of potential growth like advanced optics.
• So, what are the kinds of stocks winning from a market like High-Bandwidth Memory?
• First, the memory makers themselves are winning. This includes SK Hynix and Micron. 
• If SK Hynix is making these investments, its because of a high-confidence memory demand will reach all-time highs in the near future.
• The problem is that SK Hynix is difficult for many investors to buy unless you have an account with a broker like Interactive Brokers.
• If you’re looking for another company benefitting from the growth in HBM, you may want to consider Micron, which is an American company.
• Micron grew sales 81.5% in the first quarter, and is projected to grow sales by 51% in 2025. That’s a higher growth rate than NVIDIA, for perspective.
• It’s also worth noting that HBM supplies are effectively sold out through 2025, which is why SK Hynix is rushing to get more factories online.
• A second market that’s a winner from HBM is lithography. 
• There’s one stock that’s the biggest winner in this space: ASML.
• HMB memory has more “wafer intensity” – which effectively means more of ASML’s very expensive machines are required to make this type of memory.
• ASML already has tailwinds thanks to the growth of processors in AI, but this market just puts another tailwind at their back that’s underappreciated by investors.
• A third market that’s a winner is packaging. 
• Taiwan Semiconductor has really differentiated itself thanks to its specialty in CoWoS packaging, which is another area that’s a bottleneck for AI supply.
• Another packaging name to look at is BE Semiconductor, which is expected to see growth accelerate from negative last year to 42% sales growth in 2025.

Transcript:

Eric, the story of the investing world for the last two years has been AI and very specifically NVIDIA.

And one of the ways that investors historically have made fantastic returns on these trends is by looking at the suppliers and the components that are in the supply chains of these massive trends as they’re taking off.

So as we are looking at the next AI arms race, what are some of the big winners that investors could look forward to who are supplying these critical components and parts to the AI ecosystem?

Yeah, you know, often investors.

The original AI arms race was around GPUs and NVIDIA has been and continues to be the winner of that.

But we’re now seeing cascading other markets that are benefiting as data centers are essentially rebuilt for this age of AI.

There’s other areas like liquid cooling.

That’s been a major winner.

Power generation is an area that’s on a lot of people’s minds.

But right now, the strongest growth mark is actually high bandwidth memory, which is expected to see 78% compounded annual growth through 2027.

So now SK Hynix is not exactly a household name, but it is a leader in high bandwidth memory, thanks to kind of an early relationship with NVIDIA.

And news came out this week that plans to invest $74.6 billion in CapEx across the next five years, which is an incredible number, with its chairman saying, and I quote, the winds of AI-related change are so strong in the U.S. that there’s nothing to talk about but AI.

And he further know that the group needs to bolster its AI leadership from services to infrastructure.

So knowing that this is another market that’s really booming in AI, I wanted to talk about some of the winners from it.

Now, first up would be memory.

The first winner should be obvious.

I just talked about SK Hynix.

If they’re making this level of investment, it’s because they feel so confident about this market.

Now SK Hynix is a company it’s difficult for a lot of us investors to buy unless you have an account like interactive brokers.

The secondary option would be Micron, which actually released earnings this week, dropped a little bit.

It should be noted that they’re still forecasting incredible growth.

I believe it’s 51% revenue growth estimated from 2024, which in and of itself is amazing.

Right now, we’re running at 81%.

So this is a company growing really fast.

And they announced on their earnings that they had basically sold out their high bandwidth memory through 2025.

So it’s a company that in this high growth kind of goldmine mark of AI, they’re finding their footing in there.

The second area that’s kind of winners from an announcement like this would be semiconductor equipment.

The company I’ll focus on, it’s a company we’ve talked about several times, Austin, it’s ASML.

One of the key areas with these new memory products is they have what you’d call wafer intensity, which high wafer intensity, which leads to more demand for the lithography components that ASML sends, sells, I’m sorry.

You know, this is a company with plenty of tailwinds ads back.

People are always looking at their equipment, making things like NVIDIA chips.

They also make memory.

So this is just one more tailwind behind the stock.

The final area is packaging.

Part of high bandwidth memories is what you call, I mean, it’s just incredibly complex packaging, getting the dyes of memory interconnected to the logic or the NVIDIA chips.

Taiwan Semiconductor has done a lot of innovation in this space.

There’s also smaller companies like BE Semi, which is another company we’ve recently featured.

But one that saw negative growth throughout most of last year.

So a 9% growth in the first quarter and in the four years expected to see 42% revenue growth.

So you’re seeing these new product categories that we’re talking about the investment today of a company like SK Hynix, it’s going to flow towards incredible revenue growth.

So if investors think that they missed NVIDIA, they missed all the growth in AI.

We’re now seeing a play on these secondary markets.

Hopefully, this news and us breaking down the winners of it, hope show investors how you take some of this news and you find that next generation of companies set to profit from it.

Eric, thank you so much for putting this into context.

And it’s so easy for investors to get hung up on a company like NVIDIA or Supermicro and say, ah, I missed it.

But when we look at a trend this big, there’s so many companies and suppliers and products that are part of that.

And this is changing the dynamics of technology.

And we’ve seen how much the hyperscalers investments measured in the tens of billions or over $100 billion is in this space alone, just from a few companies.

So there’s a lot of money flowing into this space.

There are a lot of companies that provide critical components to this space, such as like SK Hynix, which was a pick that I love.

So there’s plenty of time and plenty of room for investors to still profit off of this trend without necessarily needing to jump on the bandwagon of the leading horse, which is NVIDIA.