A new Wall Street forecast suggests one of artificial intelligence’s most critical components may remain in short supply for years.
Artificial intelligence has produced clear winners.
Nvidia (NVDA) has become the poster child of the AI revolution. Cloud companies still spend billions on infrastructure. Software businesses are scrambling to bolt AI onto almost everything they sell.
But some of the biggest winners from the boom may be far less evident.
And behind every AI model is a massive quantity of memory. Without it, the most powerful chips in the industry can’t execute the calculations necessary to train and run artificial intelligence systems.
That fact is bringing Micron Technology (MU), one of the world’s biggest memory chip makers, back into focus.
A new Wall Street forecast implies the company may be riding a powerful trend that goes far beyond the current AI cycle.
More significantly, it signals a potential supply constraint that markets may be underestimating.
“Memory pricing may remain higher for longer.”
That simple assumption is getting more and more relevant for investors trying to figure out what the next leg of the AI trade will look like.
Micron is benefiting from a growing AI infrastructure challenge
Most of the talk around artificial intelligence is on graphics processing units, or GPUs.
Those chips are still important.
But more advanced memory has become just as vital.
Modern AI servers need vast amounts of dynamic random access memory, NAND flash memory and high-bandwidth memory to cope with more complicated workloads.
As AI adoption shifts from training models to inference applications, memory requirements continue to increase.
That demand is coming at a tough moment for suppliers.
Semiconductor production is not software; it can’t be scaled overnight.
It takes years to develop a new fabrication factory. Cleanroom capacity is still restricted. The time it takes to get equipment can be months long.
So the memory makers just can’t react fast enough to a sudden rise in demand.
That dynamic is helping to underpin prices across key memory markets and is providing favorable conditions for businesses such as Micron.
Related: Jim Cramer sends blunt message on Micron, Nvidia, Amazon
The supply-demand imbalance looks set to be one of the most important stories in the semiconductor industry.
Investors have focused on AI demand for much of the last two years.
They are increasingly starting to focus on AI supply.
For much of the AI boom, investors focused on companies designing processors. If memory remains constrained, future gains may increasingly flow toward companies supplying the infrastructure those processors depend on.
Wall Street believes memory prices could stay elevated longer than expected
That trend was recently noted by Wolfe Research analyst Chris Caso, who made a big upgrade to his view for Micron.
The analyst anticipates both DRAM and NAND pricing to stay relatively firm for a longer period of time, underpinned by sustained demand for AI-related products and constrained supply growth in the industry.
His prediction echoes a broader feeling that memory markets are approaching a new era.
Memory has always been one of the most cyclical sectors in the semiconductor industry. Strong pricing periods tend to create an oversupply that eventually drags prices down.
AI may be disrupting that pattern.
Unlike the past demand cycles driven by smartphones or personal computers, artificial intelligence infrastructure demands large quantities of memory and computational power.
More AI:
- Micron sits at the center of a red-hot chip rally
- IBM CEO sends blunt message on AI and quantum computing
- Anthropic CEO makes shocking admission about AI
At the same time, many customers are entering into long-term contracts with suppliers to ensure future supply.
Those contracts give manufacturers more visibility and decrease the risk of abrupt swings in demand.
MANDEL NGAN / Getty Images
What investors need to watch for next
Micron’s long-term fortunes will depend on whether AI demand continues to grow at its current pace.
High bandwidth memory, or HBM, is one of the most desired commodities in AI infrastructure. This technology allows fast data transfers between processors and memory, which is crucial for advanced AI systems.
If demand remains strong and supply remains tight, pricing could continue to rise.
That would add more support to Micron’s earnings and revenue growth.
Key takeaways
- AI servers require significantly more memory than traditional computing systems.
- Memory supply cannot expand quickly because new semiconductor facilities take years to build.
- Rising demand and constrained supply are supporting memory prices.
- High-bandwidth memory is becoming a critical part of AI infrastructure.
- Micron is emerging as a key beneficiary of long-term AI spending trends.
The greater lesson for investors may be more than any one stock.
The first wave of the AI boom benefited the firms making the chips.
The next phase might be a boon for companies that supply the infrastructure on which those chips depend.
If memory remains one of the largest bottlenecks in the sector, businesses like Micron could play an increasingly critical role in setting the pace of the AI economy’s growth.
That may be why Wall Street is getting more positive on the sector, and some experts think the memory story is only beginning.
If Wall Street is correct, the next phase of the AI trade might look different from the first. Nvidia became the face of AI, providing the GPUs powering the boom. Companies such as Micron could face a new dynamic, selling a vital component that remains tough to produce and even harder to replace.
Related: Roundhill CEO spots major shift for Micron stock