Dear reader,
The AI boom has created plenty of winners.
First, it was Nvidia (NVDA)
Then it was the power companies, utilities, and data center builders.
Now investors have found the next layer of the AI infrastructure story.
Memory.
In fact, the newest and hottest trade on Wall Street may not be AI software or robotics. It may be the memory chips that make AI possible.
The Fastest-Growing ETF in History
A newly launched ETF focused exclusively on memory stocks, the Roundhill Memory ETF (DRAM), just became the fastest-growing ETF in history. The fund accumulated $6.5 billion in assets in only 27 trading sessions, breaking the previous record held by BlackRock’s Bitcoin ETF (IBIT), which took 30 trading sessions to reach the same milestone.
The numbers are staggering.
Since launching on April 2, DRAM has surged more than 80% and recently surpassed $10 billion in assets. It now ranks among the top 10 ETFs in the United States for year-to-date inflows despite being only weeks old.
Why the Sudden Obsession with Memory?
Because every AI model requires massive amounts of memory to function.
Training and running large AI models isn’t just about having powerful GPUs. Those chips must constantly access and process enormous amounts of data. That requires advanced memory solutions such as DRAM and high-bandwidth memory (HBM).
Think of it this way.
If Nvidia’s GPUs are the engines powering the AI revolution, memory chips are the fuel system. Without enough memory, those expensive processors cannot operate at peak performance. In many cases, memory capacity is becoming just as important as raw computing power.
Demand Is Exploding — And So Are Profits
The world’s largest technology companies are spending hundreds of billions of dollars building AI infrastructure. Every new data center, every AI server rack, and every next-generation AI model requires more memory than the generation before it.
The result is a dramatic shift in demand.
AI servers can require several times more memory than traditional servers, creating a supply-demand imbalance that is pushing prices higher and boosting profits for memory manufacturers.
This is why companies such as Micron, Samsung, and SK Hynix have become some of the biggest beneficiaries of the AI infrastructure buildout. Those three companies alone account for roughly three-quarters of DRAM’s holdings.
We May Still Be in the Early Innings
Industry analysts expect AI-related memory spending to continue growing at a rapid pace over the next several years as hyperscalers race to build larger and more powerful AI systems. Several estimates suggest AI data center spending could exceed $1 trillion over the coming decade.
Of course, investors should remember that memory has historically been one of the most cyclical areas of technology. Booms are often followed by painful busts.
That doesn’t mean the opportunity is over.
Bottom Line
In my view, it highlights something I’ve been discussing for months.
The biggest winners from AI won’t necessarily be the companies building the chatbots.
Many of the biggest gains may come from the “picks and shovels” businesses supplying the infrastructure that makes AI possible.
Power.
Data centers.
Networking.
Cooling systems.
And now memory.
The market is sending a clear message.
Investors are no longer just buying AI. They’re buying everything needed to power AI.
And right now, memory chips have become one of the hottest trades on Wall Street.
Here’s to the future,
Matt McCall
Founder, NXT Wave Research
