The River Never Runs Twice
Ulrike Hoffmann-Burchardi has spent most of her professional life studying what comes next. As CIO for the Americas and Global Head of Equities at UBS Wealth Management, overseeing $7.3 trillion in assets under management, she brings to the role what few can: a PhD in financial econometrics, two and a half decades at Tudor Investment Corp building quant macro and global equity portfolios, and an almost academic obsession with technology-driven disruption. On a recent episode of the Alpha Exchange podcast 1 with host Dean Cornutt, Hoffmann-Burchardi delivers one of the more intellectually rigorous portfolio construction frameworks heard in recent months. At its center is a deceptively simple argument: most portfolios that look diversified are not.
A Lens Most Models Are Missing
The standard toolkit for portfolio construction, Hoffmann-Burchardi argues, is failing investors in plain sight. Growth, inflation, real rates, the usual macro variables, remain the primary inputs for most risk models. But a new variable has moved so far into the foreground that ignoring it has become its own form of risk.
"Some of these secular trends, structural trends have become macro factors, yet they're missing in most of the standard models," she says. "And the AI factor, in our view, needs to be a key element to stress test portfolios."
The example she offers is worth pausing on. Consider a portfolio holding a large-cap technology stock, a utility, a real estate investment trust, emerging markets, and commodities. By asset class, the pie chart looks textbook diversified. But peel back the surface and every holding shares a common thread. "They all have one common hidden factor, and that is AI." Hyperscalers profit from cloud spend. Utilities from the power demand of data centers. Commodity producers from the raw materials feeding construction. Data center REITs from the build-out itself. And emerging market indices, now dominated by semiconductor companies, from the chip supply chain underlying it all. The single variable connecting them is AI capital expenditure. If that trajectory falters, the portfolio sells off in unison.
Three Lenses, One Framework
To address this, the UBS CIO office has organized its investment process around three interlocking lenses: macro, bottom-up fundamentals, and structural trends. The structural layer, Hoffmann-Burchardi's clearest point of differentiation from the industry consensus, identifies three transformational opportunities the team calls TRIOs: artificial intelligence, electrification (which UBS labels power and resources), and longevity. AI, she notes, is the connective tissue. "AI is the one fundamental connector of all three because of course you need power and resources in order to build data centers to empower them, and then longevity is gonna be a big beneficiary of AI with drug discovery."
The framework is used not only to identify return opportunities but to stress-test risk. The team maps scenarios around each lens, asking how they interact. A rate shock, for instance, does not stay isolated. "If rates go to a point where financing for CapEx becomes an issue, of course, that means stress probably in other parts of the private markets also as some companies will be challenged with their debt load. That may mean that there is going to be larger questions about this AI build-out. The equity market is going to go down."
The Rate Sensitivity Question
The AI CapEx cycle is the defining growth trade of this era, and the question of how much rate pressure it can absorb occupies considerable attention on Wall Street. Hoffmann-Burchardi's answer is more bullish than most would expect. She notes that hyperscalers are already paying three times the going rate for electricians and twice the base electricity tariff to build their data centers. The rate elasticity of AI CapEx, in her view, is very low, and meaningfully slowing that trajectory would require far more than 25 or 50 basis points of additional pressure. The competitive logic is compelling. "Knowing that AI is such a disruptive technology... it's very difficult to believe that you would voluntarily give up to be part of this race."
She draws the analogy directly to the 1990s, a period she studied firsthand in her master's thesis on multimedia industry convergence. Of the seven dominant technology companies from that era, only two remain in today's Magnificent Seven. The stakes of being displaced are so well understood by the current players that price sensitivity to capital costs is structurally suppressed.
Playing Offense and Defense Simultaneously
The more consequential risk to the AI thesis, in Hoffmann-Burchardi's framework, is not rate-driven but rather demand-driven: a gap between what the market has priced as the monetization trajectory and what the underlying economics actually deliver. Valuations, she argues, are not the catalyst but the multiplier. "Valuations themselves are not a catalyst for market correction. It's a fundamental disappointment that's the catalyst, and valuations just tell you how much the market is likely going to fall and correct."
There is, however, meaningful early evidence on the profitability of inference, which Hoffmann-Burchardi identifies as the key forward signal. "If we can establish that inference is profitable from a gross margin and an operating margin perspective, I think the market will underwrite that and will underwrite the trajectory from here." She points to public disclosures from frontier model operators and DeepSeek's published cost analysis on GitHub as early confirmation that the math can work at scale.
To manage the inherent uncertainty, the team builds ballast directly into its structural portfolios. On days when the semiconductor and AI-enabling trade sells off, the longevity sleeve, heavily weighted to pharma and drug discovery, has shown positive return. Short-duration treasuries serve as the portfolio-wide shock absorber: "It doesn't play offense. You really play defense with short-term treasury, but it's a good place to be that gives you optionality in case there are disruptions."
The Constant Beneath All the Noise
Hoffmann-Burchardi carries one enduring lesson from Paul Tudor Jones, who gave every new Tudor trader a copy of Reminiscences of a Stock Operator. "The human emotions or the pendulum of human emotions swings between fear and greed, and that is a constant. History might not repeat exactly, but it rhymes." No amount of AI acceleration changes that. In the end, the edge she believes remains uniquely human is trust: understanding risk appetite, reading client needs, and assembling holistic portfolio solutions that no model yet replicates. "Trust is that one variable that is very difficult, I think, to replace."
5 Key Takeaways for Advisors and Investors
- AI is a hidden portfolio factor, not just a sector. A portfolio that appears diversified across asset classes may carry concentrated exposure to AI capital expenditure. Utilities, REITs, commodities, emerging markets, and large-cap tech can all move together if the AI spending trajectory slows. Stress-testing portfolios with an explicit AI factor is no longer optional.
- The rate sensitivity of AI CapEx is lower than consensus assumes. Hyperscalers are absorbing significantly elevated input costs for labor and electricity without blinking. A meaningful deceleration in AI build-out likely requires a far larger rate shock than current market debate suggests. Supply-side bottlenecks (permits, transformers, power generation) are a nearer-term constraint than financing costs.
- Inference profitability is the key signal to watch. The bridge between AI investment and sustainable market valuations runs through the profitability of inference workloads. Early public evidence suggests the economics are workable. If that narrative weakens, valuation multiples across the AI complex become highly vulnerable.
- Structural trends require dedicated portfolio sleeves, not passive exposure. The TRIOs framework (AI, power and resources, longevity) works because it enables dynamic repositioning within each theme. Playing offense in semiconductors while holding pharma as a ballast against AI drawdown days is a structural advantage passive allocation cannot replicate.
- Short-duration fixed income remains the most robust portfolio anchor. Across every risk scenario, whether rate-driven, AI-driven, or geopolitically triggered, short-duration treasuries provide optionality without directional bet. In an environment where the long end of the bond market carries meaningful uncertainty, the cost of playing defense at the short end is low and the optionality is real.
Footnote:
1 “Ulrike Hoffmann-Burchardi, Chief Investment Officer Americas and Head of Global Equities, Wealth Management, UBS. – Alpha Exchange." 29 Sept. 2026, www.axpod.com/podcast/ulrike-hoffmann-burchardi-chief-investment-officer-americas-and-head-of-global-equities-wealth-management-ubs.
