Better Artificial Intelligence Stock: Arm Holdings vs. ASML

Arm reported fiscal 2026 revenue of $4.9B, up 22.8% year over year, and $904M of net income, while ASML generated €32.7B of 2025 revenue and €9.6B of net income, for a stronger 29.4% net margin. ASML trades at a substantially lower valuation than Arm, at 27.9x forward P/E and 16.5x sales versus Arm's 125.0x and 57.3x, respectively. The comparison favors ASML for valuation, dividends and its EUV lithography monopoly, while Arm offers faster growth and expanding AI-data-center exposure but faces smartphone concentration, RISC-V competition and elevated stock-based compensation.

Analysis

The relevant distinction is not AI exposure but where each company sits in the capex cycle. ASML monetizes foundry and memory capacity additions before end-demand shipments occur, making quarterly orders, backlog mix, and customer installation schedules the key 1-3 month variables; ARM monetizes unit shipments and chip content later in the chain, with a more direct sensitivity to handset recovery, server CPU adoption, and royalty-rate mix. A soft smartphone cycle can therefore pressure ARM even while AI infrastructure spending remains strong.

ARM's valuation leaves little tolerance for execution slippage: sustained revenue growth alone is insufficient if higher-value royalty penetration, data-center wins, and operating leverage do not arrive fast enough to offset ongoing dilution from stock compensation. RISC-V is unlikely to displace ARM in premium mobile overnight, but it can cap long-run royalty pricing and gain share first in automotive, IoT, and China-linked designs. ASML's lower multiple better embeds its cyclical and China-export risk, while its installed-base service revenue and process complexity create a more durable moat than ARM's software/ecosystem switching costs.

Consensus may be over-penalizing ASML for near-term China restrictions while underestimating the ability of TSMC, Samsung, Intel, and leading memory vendors to redirect constrained capex toward leading-edge logic and high-bandwidth memory. The counter-risk is timing: semiconductor equipment orders can remain weak for several quarters before utilization rates force new capacity commitments. For ARM, the contrarian concern is that AI enthusiasm has converted a licensing business with material consumer-device exposure into a quasi-infrastructure valuation, creating asymmetric downside on even modest guidance disappointment.

Key Decisions for Investors

• Prefer ASML over ARM on a 6-18 month horizon; initiate or add on semiconductor-equipment order-cycle weakness rather than chase AI headlines. Thesis is a rerating from execution visibility and advanced-node capacity demand; invalidate if leading-edge customer capex is cut materially or export controls broaden to service/support of installed China tools.
• Express relative value through long ASML / short ARM in equal beta-adjusted notional for 3-6 months. The spread benefits if ARM's premium compresses on royalty-growth or dilution concerns while ASML bookings stabilize; stop out if ARM raises forward royalty or data-center guidance enough to support a durable earnings-estimate step-up.
• Do not add directional ARM exposure ahead of the next earnings release without confirming royalty revenue growth, net share count growth, and non-smartphone license conversion. A beat driven primarily by upfront license revenue rather than recurring royalties should be treated as lower-quality and not thesis-confirming.
• Use NVDA and TSM as read-through alerts rather than direct substitutes: sustained leading-edge GPU and foundry demand supports ASML's medium-term shipment outlook, whereas weaker global handset unit forecasts or Qualcomm/MediaTek commentary would be a more immediate warning for ARM.