Picks-and-shovels rally

Investors spent the session buying the companies that sell the pipes, drills, and concrete for artificial intelligence, not just the companies selling the chatbot itself. Micron Technology, Western Digital, ASML, and Applied Materials all jumped in the same stretch of trading as money rotated into the hardware underneath data centers. (cnbc.com) (sherwood.news) (youtube.com) That trade has an old nickname from gold rushes: picks and shovels. In chip markets, the “picks” are memory chips, storage drives, lithography machines, and wafer tools that every artificial intelligence buildout needs before a single model answers a question. (cnbc.com) (ourbrand.asml.com) Micron sits near the front of that line because artificial intelligence servers use unusually large amounts of memory. Micron said on March 18, 2026 that fiscal second-quarter revenue reached $23.86 billion, nearly triple a year earlier, as demand for memory used in data centers kept climbing. (cnbc.com) (marketbeat.com) Micron’s most prized product is high-bandwidth memory, which is a stack of memory chips built to sit right next to an artificial intelligence processor so data moves faster and with less wasted power. By early 2026, Micron had already committed its full 2026 high-bandwidth memory output to customers, which gave traders a simple message: demand is arriving faster than supply. (marketbeat.com) (futurumgroup.com) Western Digital is the slower, bulkier part of the same buildout. Artificial intelligence systems generate huge piles of training data, checkpoint files, logs, and archived outputs, and Western Digital said in its fiscal first quarter 2026 results that cloud storage demand was driving a strong environment for its business. (westerndigital.com) ASML sells the machine that makes the most advanced chips possible. Its lithography systems project circuit patterns onto silicon wafers, and ASML told investors in January 2026 that advanced logic and memory will require a higher number of critical lithography steps as artificial intelligence pushes chip complexity higher. (ourbrand.asml.com) Applied Materials comes one step earlier in the factory. It sells the deposition, etch, inspection, and packaging equipment used to build and connect those chips, and in March 2026 it announced a new partnership with Micron to work on next-generation dynamic random-access memory, high-bandwidth memory, and NAND storage for artificial intelligence systems. (evertiq.com) (techpowerup.com) The reason these stocks can rise together is that an artificial intelligence data center is a chain, not a single product. If cloud companies order more graphics processors, they also need more memory beside them, more storage behind them, and more factory gear upstream to keep the whole supply line moving. (cnbc.com) (futurumgroup.com) That is why traders often buy the hardware layer when they think artificial intelligence spending is broadening out. A chatbot company can lose users, but every company trying to build a warehouse-sized computing cluster still needs memory, storage, and chipmaking equipment before the software race even starts. (sherwood.news) (ourbrand.asml.com)