Market Snapshot: Chips & Memory

Chip stocks and memory names are trading near records as investors push the artificial-intelligence buildout beyond graphics processors and into memory and manufacturing. (finance.yahoo.com) On April 23, Nvidia traded around $202.50 in premarket, Advanced Micro Devices around $304.09, Micron around $478.52, and Taiwan Semiconductor Manufacturing Co. around $383.43 in morning trading, according to Yahoo Finance and Google Finance data. (finance.yahoo.com) Stifel raised its Advanced Micro Devices price target to $320 on April 20 while keeping a Buy rating, and Barclays lifted its Taiwan Semiconductor target to $470 on April 21 after the company’s earnings report. (benzinga.com, finance.yahoo.com) Artificial-intelligence systems need three layers of hardware: the processor that does the math, the memory that keeps data close to the processor, and the factory that builds and packages the chips. Nvidia and AMD sell the processors, Micron sells high-bandwidth memory, and TSMC manufactures and packages many of the most advanced chips. (investors.micron.com, finance.yahoo.com) That hardware chain has tightened in 2026. TSMC said last week that it now expects full-year revenue growth of more than 30% in U.S. dollar terms, up from a prior forecast of close to 30%, and Reuters reported on April 22 that the company plans an Arizona packaging plant by 2029 as advanced packaging remains a bottleneck. (bloomberg.com, finance.yahoo.com) Micron has been making the same case from the memory side. The company said in March that it had begun volume shipments of HBM4, a stacked memory product designed for Nvidia’s Vera Rubin platform, after saying in December that it had completed agreements on price and volume for its entire calendar 2026 HBM supply. (investors.micron.com, investors.micron.com) The market’s bet is that spending on large language models and data centers is no longer a one-company story. As more systems ship as full racks rather than single chips, suppliers of memory and packaging are getting valued as direct artificial-intelligence infrastructure plays, not just component vendors. (finance.yahoo.com, investors.micron.com) There are still limits to that trade. TSMC said 3-nanometer capacity used for artificial-intelligence chips remains very tight, and Micron’s long-term supply agreements show how quickly high-bandwidth memory can become constrained when cloud customers lock up future output. (investing.com, finance.yahoo.com) For now, the tape is treating chips, memory and packaging as one linked trade. The closer artificial-intelligence demand moves to physical bottlenecks in factories and memory stacks, the more investors are paying up for every layer of the supply chain. (finance.yahoo.com, investors.micron.com)