In the first half of the year, memory was the hottest AI trade. Hyperscalers poured billions of dollars into data centers where high-bandwidth memory was one of the tightest bottlenecks in the supply chain. Most of the memory stocks rallied as a result, though the returns have been muted in the second half of the year so far, accompanied by high volatility.
Some on Wall Street say memory makers are sold out till 2030. Others say only till 2028. While the exact status of the industry can be debated, there’s little doubt about the willingness of stock market investors to pay a higher multiple for memory stocks. This debate has been done to death, and memory makers continue to trade at extremely low forward earnings multiples, implying that the earnings cycle has peaked.
So what’s the truth? Is it the end of the boom? Or is the structural shift in the memory business finally here? If you would rather avoid the high-risk nature of AI stocks, particularly memory and storage, you could look at companies that are doing the exact opposite, i.e., bringing in healthy cash flows and carrying low debt in our list of 10 Stocks With Low Debt and Strong Free Cash Flow.
But if you want exposure to the best memory stocks, read on. In my quest for the answer to the above questions, I reached out to a person who regularly interviews industry experts to find out answers to questions like these. I asked a famous Substack and X analyst, P Equity Research, whether ‘this time was really different’. Here’s what he had to say:
There are essentially 3 important conditions these memory makers set in their LTAs: volume, price, duration. They are all negotiating LTAs to fix volume, a price floor, and duration of contract to get rid of past boom and bust cycles in the industry. As long as the Capex cycle holds strong, yes, they are sold out for 2028 and will be potentially beyond that.
If companies have indeed fixed the boom-and-bust nature of the business through LTAs, the next question is whether extra capacity from both US and Chinese memory makers reduces, or even entirely eats up, the margins. The answer to that depends more on the AI capex progress rather than the new supply. As long as companies continue to invest heavily in AI infrastructure, the new capacity should not cause margins to collapse. This is, however, a fear that has already caused the market to go down, leaving some AI infrastructure stocks down over 70% from their all-time highs. P Equity Research also pointed out an added risk:
I would expect spot prices to start sliding from highs over the years, which is completely normal. Will supply of memory remain tight? Yes. Will prices be just as high? I highly doubt it. More supply will increase the bargaining power of the customers by a little, and allow them to renegotiate their LTAs after the current ones expire.
While these risks are real and quite clearly affecting the market, I would like to point out that expecting new entrants to simply add capacity is easier said than done. The only competition is likely to come from China’s CXMT and YMTC, so I put that question to P Equity Research, who had an interesting take on the history of the memory business:
Over the decades, the players in memory have essentially consolidated from around 20 in the 1980s to only 3 today that are well known and in business for a long time, and then came CXMT and YMTC. Why would someone want to enter a business that is historically unprofitable and hard to manage? I would say this is a natural barrier to entry that the memory business has. If someone wants to enter the industry now, it would be too late, in my opinion. CXMT and YMTC have entered due to margins. It’s natural for high-margin businesses to lure competitors, but this is not like the GPU market where Nvidia has high-margin chips and 10 other competitors show up. It’s risky to enter the memory business, and people know that.
So how are billionaire investors treating memory stocks with so much negativity surrounding the business? That’s exactly what I wanted to find out with my article, 10 Best Memory Stocks to Buy According to Billionaires.
Methodology
To come up with my article, 10 Best Memory Stocks to Buy According to Billionaires, I first looked up multiple memory ETFs to list all memory stocks. In this list, I include stocks that derive at least 30% of their revenue from memory-related products and services. My list also bundles memory and storage stocks together. That is because some storage types also utilize memory chips and, more importantly, they face the exact same cyclicality that memory stocks are notorious for. They are listed in ascending order of the number of billionaires holding them in their portfolios.
10. Penguin Solutions Inc. (NASDAQ:PENG)
No. of Billionaire Holders: 13
Penguin Solutions Inc. (NASDAQ:PENG) is reshaping its business to put greater emphasis on AI infrastructure and memory solutions. As part of that strategy, the company has agreed to sell its Cree LED business to private equity firm Transom Capital. This is an all-cash deal and is expected to close in the first quarter of 2027. This business accounts for around 10% of Penguin’s projected fiscal 2027 revenue. Its exit will allow PENG to shift more resources toward AI data centers and memory solutions. These are the markets where it sees stronger growth opportunities.
The strategic shift comes as Penguin’s business is already picking up speed. The company generated $1.73 billion in fiscal 2026 revenue, up 26% from the previous year. Growth accelerated to 58% in the second half of 2026. Its integrated memory contributed $924 million in annual revenue, nearly twice the $464 million reported last year. Management has raised its fiscal 2027 revenue outlook as well. The company now projects around $2.43 billion at the midpoint. That would represent growth of around 40%. The stronger results give investors another reason to pay attention as Penguin puts more emphasis on AI infrastructure and memory. Previously, the stock was featured among our 10 Worst Artificial Intelligence Stocks Under $30 According to Short Sellers, so it does come with its own set of risks. Explore the full list to check which other stocks were named worst AI stocks.
9. Silicon Motion Technology Corporation (NASDAQ:SIMO)
No. of Billionaire Holders: 16
Demand for enterprise and AI storage is growing, and that is making Silicon Motion’s technology more relevant. The company designs NAND flash controllers used in SSDs, data centers, smartphones, and other storage devices. In the third quarter of fiscal 2026, revenue jumped 131% year over year to $560 million. Non-GAAP earnings per ADS came in at $4.35, up 333% from a year earlier. Gross margin also improved to 53.7%. The company’s Ferrari and boot drive solutions also grew more than 75% sequentially, which shows demand beyond traditional SSD controllers.
Through its MonTitan controller platform, Silicon Motion Technology Corporation (NASDAQ:SIMO) is expanding in enterprise storage. It has finalized the design for its next-generation 4nm PCIe 6 controller. The chip targets hyperscalers and cloud service providers, and sampling is expected to start in early 2027. Management guided fourth-quarter revenue of $588 million to $616 million. Moreover, the company said strong demand and its growing product portfolio should support further growth. So far in 2026, the stock performed well enough to secure a spot in our list of 10 Best Performing Growth Stocks So Far in 2026.
On the ownership side, billionaire Abigail Johnson’s FMR LLC reported a position of 2.28 million shares at the end of the second quarter of 2026.
8. Rambus Inc. (NASDAQ:RMBS)
No. of Billionaire Holders: 17
Rising demand for high-speed memory in AI servers and data centers is creating growth opportunities for Rambus Inc. (NASDAQ:RMBS). The company reported record third-quarter 2026 revenue of $218 million, representing 22% annual growth. Non-GAAP earnings per share climbed to $0.84 from $0.62 in the same period last year. Product revenue grew 22% to $114 million, supported by strong demand for DDR5 memory-interface chips and growing adoption of its newer products.
The company’s future growth prospects also look strong. It has secured design wins for next-generation high-bandwidth memory (HBM) and started sampling its sixth-generation DDR5 register clock driver, which supports speeds of up to 9600 MT/s. For the fourth quarter, management expects revenue to reach $227 million to $233 million, which would mark another record. The main challenge is supply. Management expects tight conditions to continue into 2027, potentially limiting how quickly Rambus can meet growing demand.
Billionaire investor Stanley Freeman Druckenmiller also showed interest through Duquesne Family Office. The firm reported a 173,000-share stake worth about $23 million at the end of the second quarter of 2026.
