Chipmakers powered Wall Street's June 30 session, with AMD surging 7.7%, Intel climbing 6% and SanDisk jumping 10.9% as investors piled into companies that build and supply the hardware behind the artificial intelligence boom rather than the hyperscalers spending on it.
The move marks a shift in how Wall Street is playing the AI trade. For much of the past two years, the biggest winners were the cloud giants — Microsoft, Amazon, Alphabet, Meta — pouring hundreds of billions into data centers. Now, with those spending plans largely priced in, traders are chasing the suppliers underneath: chipmakers, memory producers and interconnect specialists.
What happened
Nvidia added 2.6% on Tuesday, a modest gain by its own standards but enough to keep the AI bellwether near record territory. AMD's 7.7% jump and Intel's 6% advance stood out, while SanDisk's 10.9% surge reflected renewed appetite for memory and storage names tied to AI server buildouts. Marvell Technology rose 7.3% after UBS raised its price target to $340 from $230 and reiterated a Buy rating, citing the company's leading position in Compute Express Link (CXL) technology — an interconnect standard UBS expects to underpin a $7 billion to $10 billion market for AI server racks by 2030.
CNBC's Jim Cramer told viewers Wall Street is now rewarding AI suppliers — naming Micron, Intel, Marvell, AMD and SanDisk specifically — over the hyperscalers that have dominated the AI narrative since 2023. Intel, trading near $140 after a roughly 250% rally off its lows, is drawing fresh attention on questions of whether an Apple foundry deal and progress on its 18A-P manufacturing process could push shares toward $160. Marvell, meanwhile, has climbed more than 260% at points this year since Nvidia CEO Jensen Huang called it a future "trillion-dollar company" at Computex in early June.
Why it matters
The rotation into suppliers suggests investors believe the AI capital-spending cycle still has multiple years to run, and that companies selling picks and shovels — chips, memory, interconnects — carry less valuation risk than hyperscalers whose AI returns are harder to measure quarter to quarter. UBS's Marvell upgrade is a case in point: the firm raised its 2027 revenue estimate to $16.8 billion and its 2028 estimate to $23.9 billion, largely on CXL demand tied to two major U.S. hyperscalers building next-generation server racks.