Artificial intelligence-linked stocks have outpaced the broader market over the past year, with several companies in the semiconductor and networking supply chain posting extraordinary gains. Shares of memory and optical networking suppliers that benefit from AI data-center buildouts—including Sandisk, Micron, and Lumentum—have surged as investors priced in a sustained tightening between supply and demand for key components.
While momentum has already been strong, the underlying thesis for continued gains rests on whether memory capacity remains constrained and whether the shift toward optical communications keeps expanding at pace.
Key takeaways
- Sandisk shares gained about 2,900% over the past year (as of the end of August), helped by higher NAND pricing and stronger margins driven by an AI-linked SSD data trend.
- Micron shares rose nearly 700% over the past year (as of the end of August), supported by DRAM demand linked to high-bandwidth memory and limited supply capacity.
- Lumentum shares climbed about 550% over the past year (as of the end of August), benefiting from the move from copper-based connectivity to optical networks in AI data centers.
- Catalyst: A mix of supply constraints, longer-term customer agreements, and sustained AI infrastructure spending themes.
- Implication: If contracted capacity and pricing power persist through the end of the decade, earnings support could remain resilient despite the run-up in share prices.
What fueled the rally in AI supply-chain stocks
The common thread across the three companies is a shift in demand patterns tied to AI compute and storage needs. The article highlights that following the NAND market downturn after the pandemic—when demand was pulled forward for consumer electronics and computers—major memory manufacturers cut NAND capacity and redirected production toward DRAM and advanced memory configurations.
As AI chip ecosystems expanded, demand for high-capacity, flash-based storage tied to AI training workloads began to rise. At the same time, the production bottlenecks for advanced memory components used in AI accelerators have kept supply tight, supporting both revenue growth and margin expansion.
Sandisk: NAND pricing power plus longer-term contracts
Sandisk shares rose roughly 2,900% over the past year (as of the end of August), driven by a supply-demand imbalance that pushed NAND prices higher. According to the article, Sandisk is positioned as a pure-play NAND flash manufacturer, and the company has benefited from that pricing strength as well as improving profitability, with gross margin expansion over the same period.
A key issue for investors is whether the rally reflects a temporary cycle or the start of more durable demand. The article argues durability has improved as Sandisk increasingly signs longer-term agreements. It said Sandisk’s first three deals are worth a minimum of $42 billion and that the company has five agreements covering up to five years. The contracts are described as covering roughly one-third of expected fiscal 2027 capacity, with a stated ambition to raise that to over 50%.
For longer-range earnings visibility, the article cites Sandisk’s projection of revenue growth in the mid-to-high teens between fiscal 2028 and fiscal 2030 while keeping adjusted gross margin around 80%. It also notes that the stock trades at a forward price-to-earnings ratio of 7.
Investor interpretation here centers on whether higher NAND prices and contracted capacity can persist as AI training and storage needs evolve—and whether management can protect margins through demand shifts later in the decade.
Micron: constrained DRAM supply tied to high-bandwidth memory
Micron shares rose nearly 700% over the past year (as of the end of August), according to the article. Micron is described as one of the “big three” memory manufacturers, with the company’s revenue mix in the prior quarter attributed to 76% DRAM and 24% NAND.
The same supply-demand logic that supported Sandisk is applied to Micron’s DRAM exposure, the article says—particularly through demand for high-bandwidth memory (HBM), which scales with AI chip adoption. The investment angle depends on whether HBM constraints are structural.
The article highlights three constraints that keep HBM supply limited: advanced layers of HBM rely on EUV lithography equipment used for logic chips like GPUs; ASML is identified as the only supplier of these machines; and HBM requires up to three times the wafer capacity of regular DRAM, limiting how much ordinary DRAM capacity remains available.
Those dynamics, it argues, have helped send DRAM prices higher. The article also states Micron has forward visibility through longer-term contracts and notes the stock trades at a forward P/E of 6. If the memory “supercycle” extends into 2030 and beyond, the piece suggests there may be room for further upside, though it is ultimately contingent on continued AI-driven demand and manufacturing constraints.
Lumentum: optical networking replaces copper in AI data centers
Lumentum shares gained about 550% over the past year (as of the end of August), the article says, with the main driver tied to the shift in AI data-center connectivity from copper cabling to optical networks.
According to the article, Lumentum makes high-power indium phosphide (InP) lasers used for high-speed optical data transmission. It claims the company can produce these lasers at scale and holds up to a 60% market share for advanced lasers. It also lists exposure to optical circuit switches (OCS) and co-packaged optics (CPO) markets.
The piece attributes Lumentum’s competitive position to manufacturing complexity and execution scale. It says InP laser design and production require specialized semiconductor fabrication and overcoming significant engineering challenges. It adds that the company holds over 1,000 patents and benefits from decades of yield and cost advantages. After components are certified for specific AI platform deployments, the article argues switching risk to competitors is lower.
With the migration to optical networks described as just beginning, the article frames Lumentum as relatively early in a longer buildout cycle, which investors may view as supportive if data-center capex remains focused on higher-speed connectivity.
What to watch next
For AI-linked memory and optical supply chains, investors will likely focus on whether contractual commitments convert into sustained pricing power, and whether manufacturing constraints ease faster than expected. Future catalysts to monitor include memory demand signals tied to AI spending, updates on longer-term agreement rollouts, and broader macro factors that influence rates and technology capex decisions—along with upcoming company earnings and any related guidance that could confirm or challenge the durability of margins and supply discipline.
