Artificial intelligence has become, in the words of AllianceBernstein strategist Inigo Fraser-Jenkins, "all-consuming in terms of investors' attention." In a September 2026 research report1 titled "The Everything Trade," Fraser-Jenkins and co-contributors Alla Harmsworth, Robertas Stancikas and Maureen Hughes lay out a detailed, multi-dimensional case for why AI's insatiable demand for capital is not only reshaping individual asset classes but actively eroding the diversification logic that has governed multi-asset portfolio construction for decades. This is not a bearish call. It is a structural warning about portfolio robustness.
An Earnings Bubble, Not a Valuation Bubble
The equity narrative, Fraser-Jenkins argues, is fundamentally an earnings story. A 12-month forward price-to-earnings ratio of 19.4x for the US market is elevated but not alarming. What is alarming is the Shiller PE at 41x, which reveals something specific: this is "an earnings bubble rather than a valuation bubble." Whether current earnings trajectories are justified depends entirely on which AI productivity path materializes. Fraser-Jenkins models three scenarios, ranging from Daron Acemoglu's cautious 0.1% per annum productivity uplift to a techno-optimist view of 2.5%. The midrange scenario of 0.9% aligns with the McKinsey 2025 study and, critically, is broadly consistent with the combined hyperscaler cloud revenue backlog, which has now surpassed $2 trillion. That backlog effectively rules out the low-productivity scenario. The low path looks increasingly implausible. The open question is how far along the optimistic path the market is already priced.
The Financing Fault Line
The capital demands of the AI buildout are staggering in historical terms. Fraser-Jenkins places current AI capex alongside the UK and US railway booms, US electrification and the post-WWII infrastructure surge, and concludes that the AI buildout may represent the most intense phase of capital expenditure ever recorded. Critically, hyperscaler free cash flow turned negative this year for the first time. That pivot from self-funded growth to external financing has changed the composition of credit markets in ways that matter for portfolio design.
Technology's share of the US investment-grade bond index is projected to rise from 2.7% in 2025 to 8.8% by 2030. For passive credit investors, that shift introduces a growing source of correlation with equity markets at the exact moment when diversification is needed most. The report is explicit: this makes "diversification harder to come by." Within credit, the case for active over passive exposure strengthens considerably.
The Dollar, Flows and the Defensive Problem
Foreign capital inflows into US assets have reached proportions last seen during the TMT bubble and the pre-GFC era. Of the $5.2 trillion increase in inflows over the last two years, nearly 23% has gone into US equities rather than bonds. The implication is significant. Foreign demand for the dollar is now increasingly tethered to equity market performance, not sovereign bond flows. If the AI trade stalls, the dollar may offer far less protection than historical patterns would suggest. As Fraser-Jenkins states plainly, the dollar "would be less defensive in an AI sell-off."
Building Robustness Around the Core
The recommended response is not to abandon the AI trade but to build durability around it. Fraser-Jenkins maintains a strategic overweight to equities and to US equities specifically. Around that core, the report identifies several potentially defensive positions:
Healthcare trades at a substantial discount to its 30-year relative valuation average and has no structural connection to AI cycle risk. Energy has gone from a market darling to a deeply unloved sector, now carrying a negative beta to the broader index and a sharply negative correlation to high-yield credit since Q1 2026. That makes it a credible cross-asset diversifier. The yen, while difficult to trade tactically, has suffered from the same global capital flows into US AI assets that make it an intriguing mean-reversion candidate in the event of a reversal. Gold remains a long-standing conviction at AB. And value, particularly as measured by free cash flow yield, remains below its 35-year average discount despite recent performance.
Five Key Takeaways for Advisors and Investors
- AI is not just an equity story. It is a cross-asset structural force creating correlations where diversification used to live.
- Passive credit exposure is increasingly problematic. Rising hyperscaler weight in IG indices ties credit returns to the very risk advisors may be trying to hedge.
- The dollar may not protect in an AI drawdown. Foreign equity inflows, not bond flows, are now the primary prop for USD demand.
- Healthcare and energy offer uncorrelated equity exposure at reasonable valuations relative to history.
- Gold and the yen merit serious portfolio consideration as structural hedges in an environment where traditional diversifiers are compromised.
Footnote:
1 Fraser-Jenkins, Inigo, Alla Harmsworth, Robertas Stancikas, and Maureen Hughes. "The Everything Trade." AllianceBernstein, September 2026, https://www.alliancebernstein.com/content/dam/global/insights/insights-whitepapers/the-everything-trade.pdf. Accessed 25 September 2026.
