Merryn Talks Money: Bessent, Bond Yields and AI Bubble (Podcast)

Britain's bond yields are the highest in the G7 as renewed inflation pressure, rising defense costs and reluctance to cut spending intensify fiscal-market risks. The discussion also examines Scott Bessent's potential influence on US Treasuries and the yen, alongside the growing risk of a correction in the AI-led market boom.

Analysis

The actionable UK transmission is not simply higher gilt yields but a widening fiscal-risk premium versus Germany and the US. That premium raises mortgage-reset costs, restrains UK housing activity and compresses domestic cyclicals before it materially affects multinational earners; UK banks face a mixed outcome, with higher asset yields initially offset by rising impairments and weaker loan growth over the next 2-4 quarters. A persistent rise in long-end real yields would also challenge defined-benefit pension hedging and levered real-estate valuations, making UK commercial-property exposure a more acute second-order risk than the broad FTSE 100.

For US assets, any Treasury/yield-management intervention narrative should be treated as positioning-sensitive rather than a durable macro signal until auction tails, term premium and foreign demand data confirm it. The near-term vulnerability is a reversal in crowded duration-short and dollar-long positions if growth data soften; over 1-3 months, that would favor long duration and pressure financials more than it would validate a broad risk-on move. Conversely, a sustained term-premium increase would raise the discount rate on long-duration AI equities even if earnings estimates remain intact.

AI valuation risk is most acute where capex commitments are being capitalized into distant revenue assumptions. A modest slowing in hyperscaler capex guidance or evidence that inference pricing is falling faster than utilization rises could trigger multiple compression in high-beta infrastructure beneficiaries, while diversified software firms with demonstrable AI-driven seat expansion should prove relatively resilient over 6-18 months. The consensus is too focused on an immediate "AI bubble" break: the more probable path is dispersion, with financing costs and monetization evidence separating balance-sheet-rich platforms from suppliers dependent on a narrow set of customers.

Key Decisions for Investors

• Express UK fiscal-premium risk through long IGLT / short BUND on a 1-3 month horizon, sized modestly: the trade works if UK 10-year gilt-Bund spreads widen on fiscal headlines or inflation surprises; exit if a credible budget package narrows the spread by 20-25bp.
• Prefer a relative UK equity position: long defensively global FTSE 100 exposure via ISF versus short UK domestic cyclicals via EWU or selected UK housebuilder exposure. Hold through the next inflation and fiscal-policy checkpoints; the thesis is invalidated by a decisive fall in mortgage rates and upward revisions to UK consumer-demand forecasts.
• Add a tactical duration hedge with long TLT calls or call spreads dated 3-6 months rather than outright duration. Risk/reward improves if Treasury auction demand stabilizes and labor data weaken; cut if US 10-year real yields make new highs alongside firm nominal-growth revisions.
• For AI, favor a quality-dispersion pair over a sector short: long MSFT or GOOGL against short a high-multiple, single-theme AI infrastructure basket only after the next hyperscaler capex/guidance cycle confirms deceleration. Do not initiate a broad AI short without customer-concentration, backlog-conversion and capex-financing data; those are the missing variables needed to distinguish valuation risk from an earnings reset.