Magnificent Seven tech firms should watch out, says investor

Polar Capital Technology Trust warns that as AI and large language models advance, the dominance of the Magnificent Seven (including Microsoft, Nvidia and Alphabet) faces competitive risk even as hyperscalers pour ‘hundreds of billions’ annually into AI infrastructure. The £5.8bn trust — which joined the FTSE 100 in February — holds 10% of NAV in Nvidia and 8.8% in Alphabet, is underweight the Magnificent Seven, and has exited Oracle in favour of companies funding capex with equity; manager Ben Rogoff expects AI-driven infrastructure spend to rise into 2026, creating both winners and tangible risks to incumbency.

Analysis

Market-structure: AI spend rising benefits capital-intensive suppliers (NVDA, AMAT, LRCX, power/cooling operators) and hyperscaler cloud infra (MSFT, AMZN, GOOGL) while pressuring incumbents whose moats depend on ad/search or software licensing (META, ORCL). Expect winner-take-most in high-end GPUs and data-centre real-estate for the next 12–36 months; however commoditization of models will erode margins for peripheral software and ad businesses over 2–5 years.

Risk assessment: Key tail risks are antitrust actions (US/EU 12–36 months), China/Taiwan supply-chain shock (weeks–quarters), and a rapid model commoditization that collapses API pricing (12–24 months). Hidden dependencies include energy/copper supply and third-party model ecosystems (open weights enabling low-cost entrants); catalysts that could flip the trade: Nvidia earnings, a major open-model release, or regulator fines.

Trade implications: Tactical plays favor semi and infra longs (NVDA, AMAT, MSFT, AMZN) and selective shorts in legacy software/enterprise SaaS where capex is debt-funded (ORCL). Use relative-value pair trades (long NVDA or AMAT vs short ORCL), option call-spreads to express upside with defined risk, and add 3–12 month protective hedges around major earnings dates.

Contrarian angle: Consensus ties AI to Mag7; that underestimates disruption by open LLMs and specialist startups which can shave 10–30% off incumbents’ addressable pricing over 2–4 years. Historical parallel: late-90s infrastructure winners vs application losers — expect re-rating where chips/cloud win, many large software names revalue lower if they cannot monetise models.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in NVDA over the next 30 trading days, scaling in on any >8% pullback; pair with a 3‑6 month 10–25% OTM call spread to cap premium (target IRR on upside >2x, stop-loss -20%).
• Initiate a 2% long in MSFT or AMZN (cloud exposure) financed by a 1–1.5% short in ORCL (pair trade) with a 12‑month horizon; close if the relative performance gap narrows by >20% or if ORCL deleverages via equity-funded capex announcements.
• Buy 6‑month protective puts (≈20% OTM) equal to 50% notional of NVDA/MSFT longs to hedge a regulatory or supply-chain shock; alternatively implement collars (sell near-term calls 10% OTM) to monetize IV ahead of earnings in next 30–60 days.
• Overweight semiconductor equipment (AMAT, LRCX) and copper miners by +1–2% if hyperscaler capex guidance for 2025–26 is revised up by >10%; concurrently reduce legacy enterprise software exposure (ORCL, tickers with >50% debt-funded capex) by 1–2% within 30 days.