Capital in Motion: How the World's Largest Investors Are Repositioning for the AI Era

Hi Impactful Listener!

Source: Milken Institute Global Conference 2026, Beverly Hills, CA — May 4, 2026 Panelists: Marcie Frost (CEO, CalPERS), Ron O'Hanley (Chairman & CEO, State Street), Harvey Schwartz (CEO, Carlyle), Daniel Simkowitz (Co-president, Morgan Stanley) | Moderator: Sara Eisen (CNBC)
Published by: Barbara Bickham

The global economy is entering a phase of structural fragmentation — higher-for-longer interest rates, contested geopolitics, blurred lines between public and private capital, and the accelerating force of artificial intelligence. At the Milken Institute's 2026 Global Conference in Beverly Hills, four of the most consequential figures in institutional finance sat down with CNBC anchor Sara Eisen to answer a deceptively simple question: how is capital being repositioned at scale, and what does the next cycle actually look like? Their answers revealed more about the stakes of the AI transition than most quarterly earnings calls ever will.

What Is Driving Capital Allocation Decisions in 2026?

The dominant signal in global capital markets right now is artificial intelligence — and the leaders on this panel wasted no time making that clear.

"The dominant one right now on the investment side and the potential productivity side is AI," said Daniel Simkowitz, Co-president of Morgan Stanley, whose firm manages nearly $10 trillion across wealth and asset management portfolios. That framing set the tone for an hour-long discussion that ranged from CalPERS' anxiety over workforce disruption to Carlyle's proprietary data strategy, from the systemic risk debate in private credit to the financing mechanics behind trillion-dollar GPU build-outs.

The US remains the primary locus of capital deployment. Simkowitz noted that being overweight the US "has been a good trade, and it still is one you have to take extremely seriously." But he also flagged emerging alpha opportunities globally — in Japan, where corporate governance reforms have transformed equity markets; in Greater China, which is attracting incremental capital after a two-to-three-year pause; and in Brazil and the Middle East, where equity market development is accelerating. Germany, Simkowitz suggested, may be on a similar trajectory if it chooses to develop its capital markets infrastructure.

How Is AI Reshaping the Investment Thesis for Institutional Allocators?

For the world's largest pension fund, AI is simultaneously an investment opportunity and an existential concern.

Marcie Frost, CEO of California Public Employees' Retirement System (CalPERS), which manages more than $500 billion in retirement assets, described the fund's current posture as deeply focused on human capital governance. "Our board has been asking a lot of questions around AI governance and the stewardship program that we have at CalPERS," said Frost. "How are we making sure that the public companies are thinking about human capital disruption and retraining programs?"

The concern is structural, not speculative. A public pension fund like CalPERS is predicated on a continuous replacement cycle — when one worker retires, another enters the workforce and begins contributing. If AI meaningfully compresses that ratio, the downstream effects on pension solvency across the nation could be substantial. Frost articulated this clearly: "If we see that ratio start to really drop down, then that has a real impact on public pensions across the nation."

Internally, CalPERS has adopted a "human-enabled, human-in-the-loop" posture for its own AI deployments. But the harder question is evaluating AI deployment practices across its public and private portfolio companies. "It is early. It is early for us," Frost acknowledged — and her team is actively building the metrics and governance frameworks needed to assess whether portfolio companies are true AI leaders or merely AI-adjacent.

As a concrete example of the disruption she has on her mind, Frost described her commute to the conference: the abundance of Waymo autonomous vehicles on the roads and a conversation with her Uber driver about technology displacing gig economy workers. The retraining infrastructure for independent contractors, she noted, essentially doesn't exist. "Are there really retraining programs for the gig economy? Probably not. They would have to do that on their own as independent contractors."

How Is Carlyle Using AI Across Its Portfolio and Operations?

Harvey Schwartz, CEO of Carlyle, offered the panel's most operationally detailed perspective on AI adoption. With 2,500 employees at Carlyle and approximately 750,000 employees across its portfolio companies, Schwartz views AI deployment as a competitive advantage that extends across the full investment lifecycle.

"The most important decision we make is the moment that capital is deployed," Schwartz explained. "So how do we use this next set of technology to inform that investment decision — both avoiding investments, enhancing investments — and then really, how do we work with CEOs of those companies who are responsible for driving that technological change?"

Carlyle's answer is proprietary data. The firm, founded by David Rubenstein in 1987, has amassed decades of longitudinal data across private equity, credit, and real estate transactions. "What we're really trying to unlock is how to use that proprietary data so that we make better investment decisions, and also how do we use this technology to make decisions faster," Schwartz said.

Critically, Schwartz pushed back against the popular narrative that AI will primarily manifest as headcount reductions. "This is not a mission to see if we can run a company with 2,500. This is a mission to see how we run a company more productively with better outcomes and create marginal alpha." His view: AI's transformative power lies in productivity and innovation, not mass unemployment.

On the workforce question, Schwartz was similarly measured: "I'm a buyer of the productivity story... I'm not a buyer of the 'we have massive unemployment' because I'm not an advocate of that scenario." The early-stage job displacement stories are real, he said, but the more significant effect will be companies "delivering better outcomes, innovating faster, and actually being more productive."

One concern that did break through Schwartz's optimism was cybersecurity. The same exponential improvement curves driving AI's productivity promise also create compounding attack surfaces. "The things that make me nervous is the exponential compounding effect of cyber," he said, pointing specifically to the democratization of coding through AI tools — "vibe coding" — and the security risks of placing powerful capabilities in less-controlled hands. Combined with nation-state actors and critical data center infrastructure, Schwartz sees cybersecurity as the underappreciated tail risk of the AI build-out.

Carlyle's position in aerospace and defense — one of its founding investment verticals, dating back nearly 40 years — gives the firm a front-row view of how national security, large language models, and technology infrastructure are converging. "It's extraordinary the convergence around all these things," Schwartz noted.

How Should Corporations Actually Capture Value from AI?

Ron O'Hanley, Chairman and CEO of State Street, offered the panel's most structural take on how AI will reshape organizations. State Street, with 48,000 employees, operates as custodian and fund administrator for many of the world's largest institutional investors — a role that is "operationally intensive" and must be "managed at a very, very low error rate."

For O'Hanley, the technology itself is almost beside the point. "The technology, even if it froze now, is pretty incredible," he said. "It's about how it gets deployed. And in many other technological revolutions that we've seen in the past, it hasn't really changed the nature of the way the work gets done. AI is really about operating models."

The implication is profound: AI will not simply automate tasks within existing corporate functions — it will force a fundamental disaggregation and reaggregation of how work itself is organized. "I think you'll start to see functions kind of disaggregate and reaggregate around what they need and what the desired outcome is now," O'Hanley predicted. "It's not just about more efficiency... It's about how can I do the work differently."

O'Hanley also addressed the overbuilding concern directly. His answer was essentially: the incumbents have every right to win. They have the data, the client relationships, the institutional trust, and now access to the same technology as any potential disruptor. The question is whether they have the organizational will and speed to act on it. "The market is going to force a very rapid adoption of this because you're going to see these use cases that are taking off, these success stories. Investors like CalPERS are going to say, 'Why aren't you, State Street, doing this?'"

Where Is the Best AI Investment Value — Infrastructure, Chips, or Deployers?

This is the $5 trillion question, and Simkowitz gave Female VC Lab readers a useful framework.

Morgan Stanley has engaged deeply with Jensen Huang of NVIDIA — "compute equals intelligence," Simkowitz quoted, adding that intelligence "equals revenue and expenses." The firm is constructing investment baskets not just around AI infrastructure plays (data centers, chips, hyperscalers), but around what Simkowitz called "optimal deployers" — the enterprises that are most effectively integrating AI into their operations.

"The firms that deploy, so the non-LLMs or infrastructure players, but just the regular corporate... we're starting to track that and build baskets. Those who are viewed as optimal deployers versus non-optimal deployers," Simkowitz explained. The metrics: margins, ROE, CEO interview signals, and revenue acceleration data. "We've been able to create baskets and see outperformance."

On timing, Morgan Stanley's view is that we are only 10-15% through the capital deployment cycle. "We would argue we're in the very earliest stages here," Simkowitz said, framing the current moment relative to prior mega-tech cycles. The revenue acceleration rates being achieved by leading AI companies are "some of the fastest in the history of the capital markets" — yet the investable opportunity remains largely ahead of us.

The financing mechanics are worth understanding in detail. Unlike the 1990s internet build-out — which was constructing new digital infrastructure from scratch — AI is building a capability layer on top of global internet infrastructure that already exists. The primary source of capital is the world's most cash-generative companies: hyperscalers and semiconductor firms that have accumulated enormous cash flows over the past decade. Morgan Stanley recently launched its first public loan deal backed by GPUs, and has facilitated secured financing with TPUs backed by Google cloud contracts with AI firms including Anthropic. The compute market is effectively sold out, which creates collateral value supporting a new category of asset-backed finance.

"Those of us who lived through fiber and CLECs," Simkowitz noted, "this feels a lot more sturdy, both in its demand profile because we're sort of sold out on compute, and because the underlying investment is from some of the best-capitalized companies who are looking at diversification of funding."

Is Private Credit a Systemic Risk or a Healthy Evolution?

The private credit debate has intensified in 2026, with redemption requests in some semi-liquid vehicles hitting 9-10% of NAV in early 2026 — well above the 5% redemption limits that many funds have in place. The panelists pushed back firmly against the systemic narrative.

O'Hanley offered the most pointed reframing: private credit is structurally the opposite of what caused the 2008 crisis. "Banks are concentrators of risk. All that risk was concentrated, and a financial crisis turned into an economic crisis. Private credit is exactly the opposite. It's a distributor of risk." With the $1.7-2 trillion direct lending market spread across thousands of stakeholders, an elevated credit cycle would produce widely distributed losses rather than systemic cascade.

That said, O'Hanley acknowledged a genuine concern in the semi-liquid segment — roughly $200 billion of the total market, or about 2% of the combined market cap of Google and NVIDIA, as Simkowitz quantified it. The question of whether quarterly redemption rights are appropriate for five-to-seven-year loan structures is legitimate. "Should we say to individuals, 'There's a five-year term here, and redemption — actually not going to have quarterly redemption'?" The answer, O'Hanley implied, should probably be yes.

Schwartz was careful but candid about the software-specific stress. A concentrated skew toward software in both private equity and private credit portfolios collided with AI-driven disruption to software business models — and those portfolio marks are now adjusting. "Credit is credit," he said. "You just need really good underwriting, thoughtful processes." Fraud is always difficult to detect regardless of whether capital is bank-originated or private. But the economic engine remains "quite good," and he noted Carlyle is currently seeing increasing institutional interest despite the noise.

Frost brought the retail investor question home from a fiduciary perspective. CalPERS holds approximately 4% of its portfolio in private credit and 20% in private equity. The fund itself is sophisticated enough to navigate illiquidity. Her concern is about the broader democratization of private markets to retail investors who may not fully appreciate redemption constraints — especially if those investors are 50 or 60 years old with near-term liquidity needs. "I would worry about those that don't have the sophistication levels to really maneuver these complex tools."

What Is the Strategic Outlook for Institutional Capital Allocation?

Several consensus themes emerged from this panel that Unbound believes are highly actionable for institutional investors, family offices, and LPs.

AI deployment alpha will exceed infrastructure alpha over time. The infrastructure trade — data centers, GPUs, power — is crowded and early-stage valuations are already stretched. The next wave of alpha will come from identifying enterprises that most effectively translate AI capability into durable competitive advantage. This requires engagement, not just screening.

Human capital risk is an investable governance factor. Marcie Frost's insistence on tracking AI governance and workforce retraining programs at portfolio companies is ahead of the market. As AI disruption becomes visible in employment data, pension funds and long-horizon allocators will increasingly price this risk explicitly. Companies with credible human capital transition strategies will attract more institutional capital.

CalPERS' Total Portfolio Approach is a leading indicator. The fund's July 2026 shift from strategic asset allocation to a total portfolio approach — increasing active risk limits to 400 basis points and enabling cross-asset-class capital allocation decisions — signals a broader directional shift in how the world's largest allocators will compete for returns in a more complex market.

Private credit stress is real but bounded. The semi-liquid redemption pressure is a structural problem that the industry will solve through better product design, not a systemic threat. The credit cycle is turning, which will create better entry points for disciplined allocators.

Cyber is the underpriced tail risk. Harvey Schwartz's warning about the compounding effect of cyber risk — accelerated by AI-enabled attack capabilities and concentrated in critical infrastructure like data centers — deserves more attention from institutional risk frameworks than it currently receives.

The panel closed with Simkowitz articulating what may be the defining challenge of the next decade: "The social issues scare me the most." Not because he expects 20% unemployment, but because the second-order effects on demand, society, and political economy are genuinely harder to model than the technology itself. For investors and executives alike, the productive question isn't whether AI will be transformative — everyone in the room agrees that it will. The question is whether the institutions managing that transition have the governance, the data, and the organizational will to get it right.

Sources:

• Global Conference 2026 | Milken Institute

• Global finance leaders flag shifting capital flows, AI impact at Milken

• CalPERS Board Adopts Total Portfolio Approach

• Private credit's 'zero-loss fantasy' is coming to an end | CNBC

• Private Credit Markets Under Pressure: Key Risks | Sage Advisory

Subscribe to The Female VC Lab newsletter for updates, exclusive content, and to stay connected with the world of female investors.