How Long-Term Investors Balance AI Ambitions With Liquidity

Temasek, Brown and Australia’s Future Fund describe different approaches to AI exposure, private markets and preparing for shocks.

RSRohit Sipahimalaniゲスト · TemasekJDJane Dietzeゲスト · Brown University
チャンネル
Citadel Securities
公開日
2026年11月9日
動画
56:04
読了時間
11 分
原文
英語
YouTube で開く
スクロールすると動画は右下に小さくなって再生を続けます。字幕は閲覧言語に合わせて切り替わります。
閲覧言語2 言語で読めます
この記事はまだ日本語に翻訳されていないため、英語の原文を表示しています。
要約

30 秒でわかる

Temasek plans to expand AI exposure while holding more of it in liquid public securities; Brown is mapping which holdings could benefit from or be vulnerable to AI; Australia’s Future Fund prefers to assess each investment against the whole portfolio rather than set an AI target. Rohit Sipahimalani, Jane Dietze and Shikha Gupta agree that liquidity matters not just to survive a shock, but to keep investing when opportunities arise. They also debate slow private-equity distributions, private-credit risks and whether bonds can still hedge equities.

  1. 1Temasek plans more AI exposure with more liquiditySipahimalani says AI enablers were 10% of the non-Singapore portfolio and the target is 25% over five years. He wants more of that exposure in public securities to retain flexibility.
  2. 2Resilience includes money for the next opportunityGupta says Future Fund needs liquidity both to meet obligations without selling core assets and to invest through disruptions.
  3. 3Secondary sales help, but investors still need exitsDietze uses the market to manage holdings; Sipahimalani and Gupta say genuine distributions and alignment remain important.
  4. 4Private credit depends on the manager’s disciplineThe panel discusses retail redemption pressure, underwriting and the capacity to work out troubled loans through a credit cycle.
  5. 5A bond hedge depends on the kind of shockGupta sees a possible hedge against a growth shock, but not necessarily inflation; Dietze worries about changing correlations with stocks.
  6. 6Brown is testing AI inside its investment teamDietze reports faster draft memos and legal workflows, and describes a team-wide day devoted to experimenting with AI.

ポイント

  1. Temasek intends to increase AI-enabler exposure from 10% to 25% of its non-Singapore portfolio over five years, while seeking a more liquid mix.
  2. Brown’s team classifies holdings by whether AI may benefit them, leave them relatively untouched, or expose them to uncertain effects.
  3. Future Fund treats resilience as both meeting obligations without forced sales and retaining capacity to invest during disruption.
  4. Secondary sales can release capital from older private funds, but the panelists say they do not replace eventual exits and distributions.
  5. Bonds may help in a growth shock yet behave differently during an inflation shock, Gupta argues; Dietze is concerned about unstable stock–bond correlations.
  6. Private-credit manager selection includes underwriting, loan terms and the ability to work through troubled loans, not just advertised yield.
全文

チャプターごとに整えた全文を読む

文字起こしのみ
01

Position the portfolio for AI without assuming every bet wins

The three institutions use different approaches to AI exposure. They also reconsider how robotics and changing productivity could alter long-term themes.

Drew McKnight asks how long-term institutions are positioning for AI. Rohit Sipahimalani says Temasek considers its effect on the whole portfolio, including Singapore businesses it owns and manages commercially but cannot readily buy or sell.

For the rest of its portfolio, Sipahimalani says exposure to AI enablers was 10% at the end of the preceding year, equivalent to 6% of the overall portfolio. Temasek intends to raise the 10% figure to 25% over five years, while managing the risk of a fast-changing field.

He distinguishes infrastructure with long contractual cash flows, semiconductors and cloud providers, and higher-risk models and applications. At the time of the discussion, roughly half of this AI exposure was in public securities; Temasek wanted the public share above 70% within a year to preserve the ability to pivot.

Jane Dietze says Brown's investment team classifies holdings as AI beneficiaries, relatively remote from AI, or in a large uncertain middle. It reviews the portfolio line by line and revisits its judgments, seeking both exposure to opportunity and businesses that make the portfolio more resilient.

Shikha Gupta says Australia's Future Fund uses a total-portfolio approach without a fixed AI allocation target. AI appears across its credit, public and private holdings, but she stresses that transformative technology does not make every related investment attractive: valuation, risk and time horizon still matter.

Gupta expects value may also emerge as AI spreads through industries, with uneven results among companies. Keeping capital flexible gives the fund room to respond. Dietze adds that faster robotics could alter demographic assumptions, including conventional comparisons between India and Japan.

Dietze says faster autonomous driving could also affect where people live and work, and therefore real estate. The team tries to think through such possibilities over ten- and twenty-year horizons rather than assuming long-standing investment themes will remain unchanged.

Gupta says AI infrastructure spending can be inflationary in the near term, while productivity gains could eventually support growth with less inflation. The fund asks whether its portfolio could endure both the current environment and a possible later shift.

Sipahimalani argues that short-term uncertainty makes a long-term perspective more useful, not less. Temasek examines themes such as energy transition and security over a decade, but reviews its assumptions each year as technology and geopolitics change.

He says India's demographic opportunity may change in character as AI affects jobs and incomes, without necessarily disappearing. He cites healthcare and eyewear as examples where rising demand and technology could still benefit particular companies, rather than treating the whole demographic theme as one undifferentiated bet.

02

Build resilience for obligations and opportunity

The investors discuss liquidity, inflation protection and long-term risk. They describe how geopolitical shifts influence diversification.

Asked about resilience, Gupta compares a portfolio to dressing for Melbourne's changeable weather: prepare for a forecast to be wrong. Future Fund reconsidered its holdings in 2021–22 as it anticipated fragmentation, political change and technological uncertainty.

For Gupta, resilience has two parts: enough liquidity to meet capital calls and hedging needs without selling prized assets, and enough capacity to invest when opportunities arise. She says roughly a third of the portfolio changed as the fund added exposures including gold and commodities for different scenarios.

Dietze also emphasizes liquidity. Private equity distributions have been weak while commitments still require cash, so Brown needs to meet obligations and retain the ability to invest. She seeks assets with different drivers of return, though finding genuinely uncorrelated holdings is difficult.

As examples of assets less easily displaced, Dietze mentions cell towers, railroads and river barges, while acknowledging that assumptions can change. Sipahimalani defines resilience as a narrower range of long-term outcomes rather than the absence of short-term volatility.

Sipahimalani says 60–65% of Temasek's portfolio consists of relatively stable compounding businesses, including Singapore infrastructure and utilities; the remaining 35–40% takes more risk for potentially higher returns. A portfolio concentrated in equities can still fall sharply in a crisis.

The distinction for Temasek is between a temporary drawdown and permanent loss. Sipahimalani says a strong balance sheet, limited leverage at portfolio companies and sufficient liquid assets help the institution endure shocks and adjust its holdings.

Temasek wants 25% of its overall portfolio in liquid securities, he says, and is adding infrastructure, private credit and commodities alongside AI exposure. These can provide different sources of resilience in an inflationary environment.

On geopolitics, Sipahimalani says Temasek increasingly considers businesses serving large domestic markets with supply chains and technology contained within one economic sphere. He gives Indian hospitals, Chinese consumer brands and infrastructure as examples that may be less exposed to cross-border tensions.

Dietze says US–China competition, Taiwan-related risk, and uncertainty in Europe complicate Brown's decisions. She does not claim easy answers. Gupta says Future Fund has added exposure in Asia and Europe while the US remains important; diversification can reduce concentrated geopolitical risks and also create opportunities.

03

Fund new allocations and manage private-equity liquidity

Temasek outlines its planned allocation shifts. The panel discusses manager dispersion, secondary transactions and the need for actual distributions.

Sipahimalani gives planned changes for Temasek's non-Singapore portfolio: AI from 10% to 25%, core-plus infrastructure from 1% to 5%, and private credit from 2% to 5%. He says some funding could come from redirecting earlier digitalization investments, cash generated by Singapore holdings, and a marginal shift from private equity.

The private-equity shift reflects uncertainty rather than a claim that every opportunity is unattractive. He worries about AI disruption and the range of possible conditions when investments made now seek exits around 2030. Contractual infrastructure cash flows and senior, shorter-duration credit offer different risk profiles, in his view.

Dietze says Brown begins with themes and tries to gain access to strong managers, rather than setting a separate private-credit target. Private equity results vary widely between managers, and the endowment's bet has been that it can access strong ones. Slow distributions have nonetheless made the approach harder to manage.

Sipahimalani says the expected return premium of private equity over private credit has narrowed in many investors' estimates, changing the relative risk–reward calculation. He acknowledges that strong private-equity managers may still produce attractive results, but not every manager can outperform.

Gupta says each potential investment is considered against the entire Future Fund portfolio and other opportunities. Its resilience work led to more real assets, infrastructure and private credit, as well as selected hedge funds and active managers where the team believed skill could matter amid dispersion.

On the secondary market for private funds, Dietze says Brown has both invested in and used secondaries to trim holdings. More buyers can give institutions a practical way to manage portfolios when older funds have not returned cash.

Sipahimalani says selling older fund interests can free money for new vintages, but secondaries do not replace eventual exits and distributions. Gupta agrees: continuation vehicles can have a place when interests are aligned and terms transparent, not merely when they defer a problem.

04

Assess private credit and manager quality

The speakers consider redemption pressure and credit-cycle risks. They examine underwriting, workouts, incentives and the role of AI in portfolio companies.

Asked about private credit headlines and redemption pressure, Sipahimalani says the most visible problems have been in products sold to retail investors as partly liquid. He remains attracted to senior, typically shorter-duration loans, while acknowledging a credit cycle will eventually occur and that some software refinancing could be difficult.

Dietze says Brown has not seen a reason to panic in its underlying holdings but monitors them closely. Gupta also sees value in private credit, while warning that many managers have not yet worked through a credit cycle. Underwriting discipline, loan terms and a workout team matter when borrowers struggle.

The moderator, who says he works at Fortress, contrasts venture capital's reliance on a few very large winners with credit investing, where making a loan is easier than getting repaid. He expects a period of stress to reveal differences in manager skill that were less visible in benign conditions.

He adds that redemption limits in certain retail credit vehicles existed in their structures, but questions how clearly liquidity was sold and whether managers who accepted and deployed money rapidly were sufficiently selective. He thinks sorting out those problems could be healthy for the market.

Asked about distress, he says some businesses in older private-equity funds are under pressure even if he does not see a full credit cycle. He argues that some assets have not sold because owners dislike the price buyers would pay, and that persistent yields could make exits harder.

On succession at fund managers, Sipahimalani says the answer varies by firm and that some large firms have already navigated transitions. He is more focused on whether credit managers have the scale and workout capabilities to handle pressure, and on whether private-equity managers return capital rather than defer sales indefinitely.

Dietze says acquisitions of investment managers by larger platforms have occurred across Brown's portfolio. Consolidation is not automatically a problem, but she asks whether the manager's incentives remain aligned with the endowment after a sale.

Sipahimalani adds that buyout managers need to demonstrate how they can help portfolio companies adopt AI. With leverage and rising valuation multiples less dependable as sources of returns, he expects operational improvement to matter more. He treats access to expertise and the ability to implement change as manager-selection questions.

05

Audience questions on AI and fixed income

The panel weighs technology expertise against buyout experience, and considers when bonds may or may not hedge risk. They discuss the risks around AI investment and rates.

The moderator worries that some managers can continue collecting fees despite becoming effectively inactive, leaving assets with owners unable to invest in them. In an audience question about buying traditional businesses and applying AI or robotics, Sipahimalani says there may be opportunities for both venture and buyout firms.

Venture investors may understand the technology, he says, while buyout firms have experience managing acquisitions and organizational change. It is too early in his view to know which managers will combine those skills successfully, or whether new firms will emerge to do it.

An audience member asks whether fixed income can again hedge equities and private assets. Gupta says bonds look more attractive than a few years earlier and may hedge a growth shock, but not necessarily an inflation shock. She favors having several tools for different scenarios.

Dietze says Brown is concerned that the historical relationship between bonds and equities can break down, so it uses bonds less as a hedge and sometimes for income or cash management. Sipahimalani says Temasek's mandate is primarily equities, while it considers private credit for its risk-adjusted return profile.

The moderator recalls approximate figures from a discussion of government and data-center bond issuance and wonders if a slowdown in data-center financing would help fixed income by reducing supply. He explicitly says he may not have the figures right; the panel treats the broader mechanism as a question.

Sipahimalani calls a slowdown a double-edged issue. He sees two risks: whether companies earn an adequate return on rapid AI investment, and whether continued spending and higher rates affect equity valuations. He says timing and the extent of any slowdown matter.

06

Mandates, scale and how institutions invest

Each investor explains its return objectives and constraints. Portfolio size affects liquidity, transaction size and relationships with managers.

The moderator asks about each institution's mandate. Sipahimalani says Temasek seeks sustainable absolute returns above its cost of capital over a long period. Expected long-term real returns feed into Singapore's budget framework, so short-term volatility can be tolerated only in service of the long-term objective.

Dietze says Brown distributes about 5.5% of its endowment annually to the university and estimates higher-education inflation at about 3%. On those assumptions, roughly 8.5% is needed to maintain purchasing power before growing it. She describes that as a demanding hurdle.

Gupta says the Future Fund's long-term mandate is to invest for future Australians, targeting inflation plus 4.5–5.5% over time. The fund examines results over several horizons and can avoid an asset entirely rather than own it solely because it appears in a benchmark.

Scale changes what an institution can pursue. Sipahimalani describes a roughly US$400 billion Temasek portfolio, much of it in controlled Singapore holdings or private investments. Only around a quarter is readily liquid, making portfolio construction and the speed of change important.

He says typical direct transactions are often at least US$200–250 million, with a median opportunity around US$300–500 million, though early-stage checks can be smaller. Temasek caps early-stage exposure, using both venture funds and direct investments partly to understand emerging threats and opportunities for existing companies.

Gupta says Future Fund manages roughly US$300 billion and invests through external managers from a relatively small team in Australia. It combines large partnerships with smaller and emerging managers; funds of funds help provide some of that access in hedge funds.

Dietze contrasts Brown's roughly US$9.5 billion pool with the larger sovereign investors. The endowment can make much smaller commitments, including a few million dollars to get to know a venture manager, giving it a different set of options.

Sipahimalani says Temasek once emphasized one team learning across early-stage and mature investments, but scale made focus difficult. It reorganized into dedicated groups for Singapore holdings, funds and partnerships, and direct investments, while trying to preserve connections between them.

He describes taking leaders of Singapore portfolio companies to Silicon Valley and China to see AI and robotics in practice. Seeing traditional and state-owned businesses apply the technology helped make the opportunities more concrete for them, he says.

07

Where the panel sees opportunity

The closing discussion covers electrification, life sciences and practical AI use within investment teams. Gupta looks ahead to productivity in the physical economy.

For a closing positive theme, Sipahimalani highlights electrification: AI adds demand, but the opportunity also extends to other sources of energy consumption. He is also optimistic that AI, data and manufacturing capacity could support discoveries in life sciences and drug development.

Dietze focuses on using AI inside Brown's investment team. She says a draft investment memo that once took 40 hours can now be produced in 40 minutes, and describes a legal workflow that became faster and cheaper. These are her team's reported experiences, not general guarantees.

Brown held a day-long team hackathon devoted to using AI and plans another. Dietze says setting aside time let colleagues experiment with tasks and process improvements they would struggle to fit into an ordinary workday.

Gupta says she looks forward to autonomous driving personally and, as an investor, to seeing whether AI produces productivity gains in the physical economy. She believes the next five years may look very different, then the panel closes.

同じチャンネルの記事

スライド 4 枚1:04:12EN
Citadel Securities· 2026年11月12日ポッドキャスト · 2 人Benchmark the hardware conditions your production code may actually faceKris Jusiak explains why tight-loop averages mislead and how to test the same machine code across branch and cache conditions.Software engineering開発ワークフローとテストオブザーバビリティと信頼性
1:01:02EN
Citadel Securities· 2026年10月27日ポッドキャスト · 3 人Sequoia's Alfred Lin and Konstantine Buhler on AI, Investment and the Agent Economyゲスト Alfred Lin, Konstantine Buhler · ホスト Nohshad ShahAt Citadel Securities' 2026 conference, the partners discuss their investment process, model competition, computing constraints and future opportunities.Venture CapitalAI StrategyAI agents+2
スライド 4 枚18:55EN
Citadel Securities· 2026年10月26日ポッドキャスト · 4 人Jeff Maurone on Citadel Securities’ Market Analytics and TechnologyThe technology COO describes tools for market positioning, credit spreads and client questions, and the research infrastructure behind them.金融 AIインフラとデプロイInvesting+2
44:58EN
Citadel Securities· 2026年10月26日ポッドキャスト · 2 人Mario Draghi on Europe’s Growth, Energy and Security Choicesゲスト Mario Draghi · ホスト Lisa AbramowiczIn an October 2026 conversation with Lisa Abramowicz, the former ECB president discusses fiscal pressures, AI, energy and European integration.GeopoliticsEconomicsAI Adoption+2
45:34EN
Citadel Securities· 2026年10月26日ポッドキャスト · 2 人Henry Paulson on AI Risk, U.S.–China Relations and Crisis Leadershipゲスト Henry M. Paulson, Jr. · ホスト Jim EspositoThe former Treasury secretary discusses today’s economic risks and lessons from the 2008 financial crisis.LeadershipDecision MakingInvesting+2
1:01:23EN
Citadel SecuritiesポッドキャストJensen Huang on AI Factories, Digital Workers and Why the Build-Out Has Barely BegunNvidia's CEO tells a room of institutional investors how accelerated computing led to AI factories, and where agentic AI, robotics and sovereign AI go next.企業と戦略インフラとデプロイAI Strategy+3