Not Micron or SanDisk-Lam Research May Be the AI Memory Boom's Biggest Winner

Lam Research may be the less obvious way to play the AI memory squeeze

Memory-stock investors are already fighting the last trade

Over the past year, memory stocks became one of the easiest trades in AI. Micron gained 657% and SanDisk more than 2,700%, but sentiment reversed quickly, with both stocks falling 33% and 50% from their peaks. The rally got crowded, and the pullback looks like investors unwinding that consensus position.

The bigger question is whether the memory shortage is actually easing. Bears think it is; Deloitte says the memory crunch is unlikely to ease until 2029. That disconnect helps explain why the memory trade still feels unsettled.

Why Lam Research matters in this setup

Lam Research sits upstream from that fight. It sells the equipment that lets memory makers expand, and 46% of its revenue comes from memory manufacturing equipment. That matters because Lam does not depend on spot pricing alone; it also benefits when customers add capacity, increase process complexity, or invest in packaging.

Yes, Lam has already had a huge run-more than 320% over the past year. But that is part of the point. While memory investors argue about cycle timing, Lam is positioned to benefit from the capital expenditure needed to relieve the bottleneck.

The bottleneck is a mix shift toward HBM, not just a temporary tight market

AI demand is changing how memory capacity gets used

The market is still thinking in cyclical terms, but the structure has changed. Memory producers are reallocating capacity toward AI data-center HBM, with one analysis describing a structural reallocation toward HBM. Because a single HBM wafer displaces two or more conventional DRAM wafers, the same wafer output can produce less standard memory bit supply even if total wafer starts remain steady.

For Lam, that matters because higher-value memory does not just mean more demand for tools; it can also mean more complex processing and greater packaging spend.

Pricing is still pointing to scarcity

SK hynix reported roughly 30% QoQ DRAM ASP growth, along with a mid-50% QoQ NAND ASP increase, doubled enterprise SSD revenue, and more than a threefold sequential rise in 30TB-and-above enterprise SSD revenue. Those are encouraging signs for a market still dealing with tight supply.

Stronger pricing can change customer behavior. When buyers expect tighter supply and higher prices, they often secure capacity early. For equipment suppliers, that can mean continued ordering even when part of the market sounds cautious.

Lam is already translating tightness into financial results

In the June quarter, Lam reported $6.72 billion in revenue and a 52.0% non-GAAP gross margin. Those numbers suggest customers are still spending at a strong level, and that Lam is maintaining profitability even as the product mix evolves.

The forward setup also looks supportive. Lam just delivered a double beat and raised guidance, reinforcing the view that the company is still in the mid-stage of an AI-driven upswing. Management also raised its 2026 WFE forecast of $135 billion-$140 billion, which points to continued industry spending.

A simple way to frame the chain is:

• HBM pulls more capacity into higher-value memory
• Tightness can support DRAM, NAND, and enterprise SSD pricing
• Memory makers then keep investing in fabs, tooling, and packaging
• Lam captures part of that spending through equipment and packaging exposure

Bears will argue that extreme pricing eventually invites oversupply. That is possible. But the current evidence still shows scarcity feeding through to pricing and earnings rather than clear relief.

Lam Research is expensive, but the debate is about valuation framing

After a more than 320% surge over the past year and a move above the $500 billion market-cap mark, Lam Research is not an easy chart to buy. The real debate is whether the market is still valuing it like a plain memory-cycle proxy when the backdrop now includes structural reallocation toward HBM and accelerated fab construction.

Bulls see Lam as an AI-infrastructure bottleneck play. Bears see a highly advanced stock that could still be exposed to a normal cycle turn. Recent results do not settle that argument, but they do narrow it. Lam reported $6.72 billion in June-quarter revenue and a 52.0% non-GAAP gross margin, while SK hynix highlighted roughly 30% QoQ DRAM ASP growth and a mid-50% QoQ NAND ASP increase. That supports the view that the squeeze is still transmitting upward, even if it does not prove the cycle cannot reverse.

What investors should watch next

• Whether HBM capacity continues pulling away from conventional memory
• Whether ASP trends in DRAM and NAND stay firm
• Whether Lam can keep revenue and margins strong as the spending cycle continues

If those signals hold, Lam can justify a rich valuation because expectations are being validated. If they weaken, the multiple likely becomes harder to sustain.