Who Benefits From AI Spending? JP Morgan's Top Picks
Strong artificial intelligence (AI) and hyperscale demand are driving significant growth for semiconductor companies, with Nvidia (NASDAQ:NVDA), Advanced Micro Devices (NASDAQ:AMD), and Broadcom (NASDAQ:AVGO) poised to benefit from expanding data center and AI infrastructure investments.
- MRVL is trading at N/Ax earnings, in line with the market average. Check the fundamentals here
Rising capital expenditures among top U.S. hyperscalers and broader AI infrastructure projects are fueling multi-year growth opportunities for AI-focused semiconductors.
JP Morgan analyst Harlan Sur projects significant upside in data center capital expenditures, as spending expectations for the top four U.S. hyperscalers have risen sharply over the past four months.
Also Read: Nvidia, Marvell, Broadcom Bounce Back After ASML Earnings Boost Investor Confidence
The firm's IT Hardware team now forecasts 2025 data center spend to grow roughly 60% year-over-year, up from 40% previously, implying more than $100 billion in incremental investment relative to 2024.
This surge in spending outpaces the roughly $82 billion in incremental AI-related graphics processing unit (GPU), Application-Specific Integrated Circuit (ASIC), and networking revenue projected for semiconductor leaders such as Nvidia, AMD, Broadcom, and Marvell (NASDAQ:MRVL) in 2025.
Capex
The analyst noted that capital expenditure (capex) growth expectations have trended higher through 2025, mirroring trends from 2024. He anticipated further upward revisions for 2026 and 2027 as confidence in sustained hyperscaler spending grows.
Data center capex is now projected to grow by approximately 30% in 2026, representing an additional $80–85 billion in investment, which could ultimately rise above 40% year-over-year, pushing incremental spend past $110 billion.
He emphasized that focusing only on the top hyperscalers underestimates total AI infrastructure spending, as neoclouds and sovereign AI projects increasingly drive material investments.
This spending environment supports a 40–50% compound annual growth rate for the AI accelerator total addressable market, based on a $200 billion base in 2025, he added.
Sur expects hyperscaler capital expenditure intensity to remain structurally elevated over the long term, supporting higher out-year capital expenditure forecasts.
The aggregate cloud capital expenditure (capex) as a percentage of revenue for the top five U.S. hyperscalers currently stands at ~18%, with projections reaching 21% by 2026.
The analyst anticipates capital intensity will stay above historical levels, driven by incremental AI compute demand and regular refresh cycles, even if it eventually retreats from peak levels.
Sur noted near-term trends suggest capex intensity may rise further as AI infrastructure spend outpaces revenue growth, with a projected 25% intensity in 2028 potentially generating nearly $200 billion in incremental spend relative to 2026, creating material upside for AI-levered semiconductor companies.
Which Company Benefits?
The analyst identified Nvidia, AMD, Broadcom, Marvell, Micron (NASDAQ:MU), and Western Digital (NASDAQ:WDC) as top beneficiaries of strong AI server and hyperscale spending.
Nvidia's datacenter GPU business will benefit from sustained AI-related demand, he noted. AMD is positioned to capture growth via its MI400/MI500 GPU series and OpenAI supply agreements, as per Sur.
Broadcom and Marvell will capitalize on hyperscale XPU ramps, he noted. Micron's HBM3E memory supports continued market share gains, Sur added.
Price Action: Nvidia shares were up 1.07% at $181.75 at the time of publication on Thursday. The stock is approaching its 52-week high of $195.62, according to Benzinga Pro data.
Read Next:
Photo by khunkornStudio on Shutterstock