NXP Semiconductors, Broadcom among Deutsche Bank’s chip stocks for 2025

The semiconductor market has been a tale of two stories in 2024, as artificial intelligence-linked companies have seen strong growth, while others have had uneven performances. Looking to 2025, Deutsche Bank is expecting much of the same.

“Looking into 2025, we expect ex-memory semi revenue growth to accelerate to [roughly 16% year-over-year], albeit with the bifurcation theme persisting (AI up low teens, remainder of sector up [low-single-digit percentages]),” analysts at the firm wrote in a note.

“In general, after a year of relative underperformance for the SOX (~-6% vs. S&P), and with fundamental growth likely improving both in magnitude and breadth, we are more constructive on the group than at this point a year ago, albeit with selectivity remaining paramount.”

Themes to think about

Looking to next year, Deutsche Bank sees two themes emerging: companies that are “broad-based” are likely to see improved growth next year and AI-related tailwinds should persist.

Amongst the broad-based semiconductor companies, Deutsche Bank favors NXP Semiconductors (NASDAQ:NXPI) and On Semiconductor (NASDAQ:ON) on the belief that expectations for the two are “at the lower end,” there have been structural improvements and valuations are still lower.

“For the broad-based names, cyclical bottoms were largely found in 2024, so the focus in 2025 will shift to the pace of recovery,” the analysts wrote. “On this topic, we expect [year-over-year] improvement directionally, but remain conservative on the magnitude given macro, inventory, pricing headwinds.”

More to AI than just Nvidia

And while Nvidia (NVDA) has dominated the AI theme in 2024 — and to a lesser extent AMD (AMD) — Deutsche Bank likes Broadcom (NASDAQ:AVGO) and Marvell Technology (MRVL). The pair have exposure to the hyperscalers continuing to spend, as well as creating application specific integrated circuits for companies like Google (GOOG) (GOOGL) and Amazon (AMZN) with their tensor processing units and Traininum2 chips, respectively.

“On the AI front, we expect GPU-based accelerators to deliver yet another year of outsized growth (led by NVDA, and somewhat AMD), with ASICs and Networking players also contributing (AVGO, NXPI, ALAB, etc.),” the analysts added.