Various high-profile chip and networking stocks can outperform again in 2026, bucking fears of an artificial-intelligence bubble, according to J.P. Morgan.
Spending on data-center capital expenditures could rise another 50% next year, after surging an estimated 65% in 2025, J.P. Morgan analyst Harlan Sur wrote on Tuesday. The trend could benefit hardware plays across the semiconductor value chain, including makers of memory components and networking technology, he added.
Among semiconductor stocks, Sur named Broadcom, Marvell Technology, Analog Devices and Micron Technology as his top picks for 2026. His only chip-equipment pick is KLA, and his only chip-design-software pick is Synopsys.
Broadcom and Marvell both could benefit from growing adoption of application-specific integrated circuits, which rival graphics processing units for AI use cases. Broadcom could deliver at least $55 billion to $60 billion in AI revenue for fiscal 2026, according to J.P. Morgan’s estimates. Meanwhile, Marvell faces “a strong setup” into the new calendar year as its AI ramp continues and as cyclical businesses turn around.
Analog Devices occupies the No. 2 position in the market for high-performance analog, and Sur likes the company’s potential to book $500 million to $600 million in AI revenue next year on the heels of design wins.
Micron’s stock has been a big winner recently, rising 173% over the course of 2025 to date. But next year could be strong as well, according to Sur. “We believe the stock still has ample room to run in 2026, supported by pricing momentum, further margin expansion” and the likelihood that financial estimates could move higher, he wrote.
Sur’s team also likes shares of Nvidia, Lam Research, Cadence Design Systems, Applied Materials, Western Digital and Astera Labs, although those aren’t among the analysts’ top picks.
Sur said that while concerns about an AI bubble will likely continue as spending continues to trend higher, “as we look ahead, we see increasing adoption of AI across consumer and enterprise markets, coupled with rising compute intensity for AI workloads, continuing to drive significant demand for accelerated compute and related infrastructure.”
While he expects infrastructure deployments next year to be limited by supply-chain constraints, “the silver lining,” Sur said, is that shortages can potentially extend the cycle of AI spending, thereby “creating a runway of several years of growth that directly benefits the semis industry.”
Additionally, Sur said he anticipates trends in the industrial and automotive markets to improve next year.
Sur’s team is modeling for semiconductor-industry revenue to grow between 10% and 15% next year, and companies more exposed to the AI infrastructure buildout stand to gain the most, he said.
Comments