U.K. stocks higher at close of trade; Investing.com United Kingdom 100 up 1.10%

BofA highlighted top chip stocks, citing server CPU total addressable market reaching $125bn by 2030, which is constructive for select semiconductor names. The broader article was otherwise a market recap: the UK 100 rose 1.10%, with M&S up 6.64%, Rolls-Royce up 5.15%, and Antofagasta up 4.74%, while Experian fell 2.95%. Commodities were weaker, with crude oil down 5.54% to $98.60 and Brent down 6.18% to $105.17, while GBP/USD rose 0.42% to 1.35.

Analysis

The key implication is not the headline CPU TAM number itself, but the widening capital-allocation gap it creates across the AI infrastructure stack. If server CPUs can still expand into a $125bn market by 2030, the market is implicitly pricing a longer runway for general-purpose compute than the current "GPU-only" narrative suggests — which is supportive for names exposed to enterprise refresh cycles and inference-heavy deployments rather than just frontier training. That shifts the winner set toward companies with durable socket share, strong OEM/channel relationships, and pricing power in enterprise refreshes.

RELX stands out as a relative loser in this tape because the market is rotating toward hard cyclical and AI beneficiary exposure while the stock trades like a quality compounder with less optionality to the current AI capex theme. The second-order effect is that analysts will likely keep rewarding "AI adjacency" over pure information services until there is clearer evidence that data/decision workflows are monetizing AI at the same pace as semiconductor infrastructure. In that regime, RELX can underperform even if fundamentals remain intact, simply because it lacks a near-term narrative catalyst.

The contrarian risk is that the CPU TAM forecast may be too late-cycle optimistic: if edge inference, custom ASICs, or rack-scale accelerators take share faster than expected, the market could be extrapolating peak relevance for x86 server CPUs just as hyperscalers optimize away from them. That would pressure the crowded longs in the CPU-enabler basket over a 6-18 month horizon, while creating dispersion between true platform winners and incremental beneficiaries. SMCI and APP remain useful beta expressions to AI infrastructure and compute monetization, but the risk is that valuation support depends on continued broadening of spend, not just magnitude.

Near term, the more actionable setup is relative value rather than outright longs: the commodity and FX backdrop is mixed, so stock selection should dominate. The big move in oil introduces a potential inflationary impulse that could later support rate-cut expectations if growth rolls over, which would favor quality software and data franchises — but only after the market stops rewarding pure AI capex beta. For now, the path of least resistance is to own the names with the clearest exposure to enterprise/server refresh and avoid paying up for businesses where AI is a theme, not a line item.

Key Decisions for Investors

• Reduce RELX on rallies over the next 1-2 weeks: use it as a funding source for higher-beta AI infrastructure exposure; risk/reward is skewed negative if the market keeps preferring tangible AI capex beneficiaries.
• Maintain a tactical long SMCI vs. RELX pair for 1-3 months: SMCI captures the AI infrastructure spend impulse while RELX is less likely to re-rate from the current narrative; stop if AI hardware multiple compression broadens.
• Buy APP on weakness for a 3-6 month trade, but size modestly: it benefits from the monetization of AI attention and ad optimization; risk is a rotation back to hardware names if the market narrows the theme.
• If adding semiconductor exposure, prefer a basket of server-CPU/enterprise-refresh beneficiaries over pure training-beta names; the trade works best over 6-12 months if inference and refresh cycles keep broadening.
• Consider a downside hedge on crowded AI hardware beta via short-dated call spreads on SMCI into strength; payoff improves if the market starts discounting slower adoption or margin normalization.