3 Artificial Intelligence (AI) Stocks That Look Like Great Buys Right Now

The article identifies ASML, Taiwan Semiconductor and Broadcom as potential long-term AI-stock buys after recent declines: they are respectively 18%, 12% and 32% below their 52-week highs. ASML expects about €45 billion in revenue this year, up approximately 38% year over year; Taiwan Semiconductor reported revenue growth of 51% and earnings growth of 69% in its latest reported period, while Broadcom’s AI semiconductor revenue rose 221% in the period ended Aug. 2. The author argues their growth prospects and forward P/E multiples of 29, 21 and 19, respectively, make the pullbacks attractive; the article cites Iran-war concerns and the possibility of higher future interest rates as factors weighing on tech stocks.

Analysis

The key portfolio implication is not that all three names are equally attractive on a dip; it is that AI exposure spans different bottlenecks. Broadcom’s custom accelerators can take some workloads from NVIDIA, but are not a wholesale substitute for its GPU/software ecosystem. A relative-value position is therefore more defensible than treating the two as interchangeable: the thesis is share rotation at the margin, not an outright collapse in NVIDIA demand. Verify customer concentration and AI-related revenue definitions before sizing.

ASML and TSM sit upstream of that competition, but both remain exposed to the same capex cycle: if hyperscalers slow spending, orders and utilization expectations can reset before long-term AI demand changes. ASML also faces export-control and order-timing risk; TSM adds a distinct Taiwan geopolitical tail that valuation alone cannot hedge. The article’s growth and forward-multiple comparisons rely on different reporting periods and estimates, so they are not clean evidence of relative cheapness.

Near term (days), war-driven energy inflation and higher yields can pressure long-duration semiconductor multiples regardless of company execution. Over 1–3 months, watch hyperscaler capex commentary, foundry utilization/order signals, and revisions to AI revenue expectations. Over 6–18 months, accelerator substitution could benefit Broadcom while keeping TSM demand supported; ASML benefits only if that demand translates into sustained leading-edge fab investment. Contrarian risk: a broad dip may be mistaken for opportunity when consensus AI spending expectations are already embedded. No indiscriminate basket buy is warranted.

Key Decisions for Investors

• Consider a small, market-neutral long AVGO / short NVDA relative-value position only after confirming current AI revenue mix and customer concentration. The bet is incremental custom-accelerator substitution, not a view that NVIDIA loses its platform moat. Reassess if Broadcom’s AI growth or customer commentary weakens while NVIDIA demand remains resilient.
• Do not chase ASML or TSM solely because they are below recent highs. Add exposure in stages after rate-driven selling stabilizes; for TSM, keep position sizing below a comparable non-Taiwan semiconductor exposure because geopolitical gap risk is difficult to hedge.
• For the next 1–3 months, monitor hyperscaler capex guidance, ASML order commentary/export restrictions, TSM advanced-node utilization, and estimate revisions. A material capex slowdown or downward guidance revisions would falsify the dip-buy thesis; sustained investment and stable orders would support adding.
• If real yields rise further or energy prices reaccelerate on the Iran conflict, reduce gross semiconductor exposure rather than assuming company-specific strength will offset multiple compression.