For decades, the asset class pie chart has served as the visual shorthand for prudent portfolio construction. Five buckets, five colours, the geometry of safety. Ulrike Hoffmann-Burchardi, CIO for the Americas and Global Head of Equities at UBS Wealth Management, argues in a pointed new commentary that this familiar picture can now be dangerously misleading. Her case is methodical, specific, and hard to dismiss.
The Same Risk, Five Different Labels
Hoffmann-Burchardi opens with a literary provocation. She invokes the Swiss author Max Frisch, whose novels warned against reducing people to fixed identities. The analogy to investing is direct: "The same categorical thinking can be harmful in investing. AI, in particular, is challenging traditional asset class labels. Diversification by asset class can create an illusion of diversification."
To prove the point, she constructs a hypothetical portfolio that, on any conventional chart, would appear well-diversified: large-cap tech, REITs, utility stocks, industrial commodities, and emerging markets equities. Five distinct buckets. One hidden concentration. Each, she argues, has quietly converged on the same underlying driver: AI compute demand.
The mechanics are worth following closely. A megacap tech stock and a data center REIT occupy separate asset class categories, yet their fortunes are now structurally entwined. "The tech company's growth is throttled by the physical footprint of the REIT, and the REIT's cash flow depends on the tech giant's AI capital expenditure." Regulated utilities, historically priced as bond-like defensives, are no longer playing that role: "as AI data centers push up demand for electricity, some utility companies are now priced as high-growth AI infrastructure plays." Copper, filed under basic materials, is being consumed by grid expansion and specialized cooling systems for data centers. And emerging markets, perhaps the most counterintuitive entry on the list, are no longer the diversifying exposure they once appeared to be. "Over the last 12 months, just three AI semiconductor stocks generated more than 60% of the MSCI Emerging Markets Index's total performance." The key driver, she repeats with deliberate emphasis across each example: AI compute demand.
The conclusion follows cleanly. "Viewed through separate asset class lenses, the portfolio is blind to this concentration. Yet during an AI slowdown, all five of these diversified buckets would likely experience a correlated drawdown."
Scenario Thinking as the Alternative
Hoffmann-Burchardi's prescription is a shift in framework, not just a rotation in holdings. UBS CIO's investment approach, she explains, operates through three analytical lenses: macro factors including real GDP growth and real interest-rate sensitivity; structural trends, which she identifies as AI, power and resources, and longevity; and bottom-up fundamentals covering valuations and company-specific risks across public and private markets.
The practical advantage of this approach extends beyond risk identification. It enables capital to move where opportunity actually is, unconstrained by the bureaucratic logic of a full sleeve. "If a market shock misprices private credit, for example, an asset class view can impose structural limitations if the fixed-income sleeve is already full." Under a single, integrated risk budget, the marginal return of a new opportunity can be evaluated honestly against the whole portfolio, not blocked by an artificial ceiling in one category. "Capital can be shifted from liquid equities to private credit without breaking artificial asset class limits."
In short, the model becomes a balance-sheet optimization exercise rather than a box-filling one. Illiquid private business equity, concentrated stock options, and real estate holdings can be incorporated coherently into a single framework rather than shoehorned into whichever sleeve accepts them.
The Recommendation
For investors specifically concerned about AI concentration risk, Hoffmann-Burchardi recommends positioning in assets "that can compound independently of the AI evolution." She identifies two categories: quality companies with differentiated market positioning, and quantitative strategies.
Five Key Takeaways for Advisors and Investors
1 Asset class labels can obscure shared risk. Large-cap tech, REITs, utilities, copper, and EM equities may all be expressions of a single AI compute demand factor, regardless of how they are categorized.
2 Correlated drawdowns ignore category boundaries. An AI slowdown would likely pressure all five "diversified" buckets simultaneously, making conventional diversification inadequate protection.
3 Scenario analysis is a more honest framework. Macro, structural, and bottom-up lenses reveal actual risk concentrations that asset class silos conceal.
4 A single risk budget unlocks flexibility. Evaluating all positions against one integrated budget allows capital to move toward genuinely mispriced opportunities without structural interference.
5 Compounding independently of AI evolution is the goal. Quality companies with durable competitive positioning and quantitative strategies offer the clearest path to returns that do not depend on the AI buildout continuing uninterrupted.
Footnote:
Hoffmann-Burchardi, Ulrike. "Beware of the AI Factor in Your Portfolio: The Illusion of Diversification." LinkedIn Pulse, UBS Wealth Management, 2025, www.linkedin.com/pulse/beware-ai-factor-your-portfolio-illusion-ulrike-hoffmann-burchardi-r3x0e/.
