This is an edited transcript of the speech GIC CEO Lim Chow Kiat gave at the Wealth Management Institute (WMI)’s Global-Asia Family Office (GFO) Summit 2026.

Three things are asking markets for money right now, and all three want money with a long life. Governments need it to fund their promises. The artificial intelligence (AI) build-out means trillions to spend on compute, power, and grid connections. And energy systems are being rebuilt for a different pattern of demand arising from climate change. They are asking at a time when lenders demand more compensation to commit for long, especially as the range of outcomes around all three is unusually wide too. There is no shortage of capital in the world. What is shorter is patience, and patience is being repriced.

Rising yields: The good and the bad

Let me start with the bond market, because it sets the terms for everything else. The US 10-year is trading around 5.3%, and the 30-year above 5.6%1. Long rates are also rising in other major markets. We have not seen these levels in at least two decades.

Are yields rising for good reasons, or for bad ones? Good reasons would be stronger growth and more investment, with lenders asking for more because the economy can pay more. Bad reasons would be doubts about inflation, or about how much debt a government can carry. On the screen the number looks the same. In your portfolio it is not.

For all the geopolitical tension of recent years, the world economy has been more resilient than most expected, and corporate earnings have held up. The higher growth, if sustained, can pay for the higher interest rates. In addition, signs of longer-term inflation are also comforting, with contained inflation expectations.

Some of the rise in yields is also structural. For four decades, the world had more savings than it wanted to invest—ageing savers, reserve accumulation, and cautious corporates—and that surplus pushed real long-term rates down for a generation. Many of us built our portfolios in those conditions.

That balance is now inverting. The world wants to build again: compute, power, grids, defence, and supply chains that are less exposed to geoeconomic disruption. Meanwhile the pool of savings is drawn down by retirement and by deficits. A savings glut is becoming a savings shortage, and the price of long-term money is being reset by forces that outlast any one central bank meeting.

For most of the past two decades, the large advanced economies borrowed at real rates below their growth rates, and debt ratios fell with little effort. Those were unusual conditions. Judged against the real rates the market is pricing five years out, rates now sit above growth in the US, the euro area, and the UK.

Stabilising debt in this environment means governments must take in more than they spend, before interest costs. Most are doing the opposite. The IMF puts global public debt at just under 94% of GDP last year and projects it to cross 100% by 20292. Governments refinance over many years, though, and much of what they owe was borrowed cheaply. So this builds gradually.

A real investment boom can push rates above growth, and so can a loss of confidence. What makes the second dangerous is that it feeds on itself. Lenders who cannot forecast inflation charge more to lend long, and that premium buys the economy nothing. Lenders who doubt a government’s finances charge more again, and that cost is what makes the debt harder to manage.

If what we are building in AI, energy, and supply chains genuinely raises productivity, growth catches up and the gap closes. If it does not, we will have paid more for money without earning more from it. So I would watch not just where rates go, but what is driving them: growth, inflation, and what lenders charge on top for the uncertainty around both. Knowing the underlying drivers matters a great deal for capital allocation.

AI: Who pays for the build-out

AI is the clearest test of whether the spending pays off. Two years ago, investors were still asking whether AI would create real economic value. The question now is who captures that value, and who is paying for the build-out.

Increasingly, the answer is the bond market. Hyperscaler capex is approaching 2% of US GDP3, and more of it is funded with debt. S&P expects the six largest to run negative free cash flow through 20274. The Dallas Fed estimates that AI-related investment-grade issuance could add US$360bn of ten-year-equivalent duration this year5, roughly an eighth of what the US Treasury itself supplies. The build-out is competing for the same long-duration savings that governments are also drawing on. AI has become a credit story, not only an equity one. If you hold investment-grade credit, you may hold more technology risk than you used to.

Capital is not the only constraint: power, grid connections, transformers, land, water, skilled trades, and increasingly, consent. Communities are pushing back on where data centres are built and what they draw from the local grid. Bottlenecks like these are also opportunities: semiconductors, power generation, and grid infrastructure all need long-term capital.

We use two frameworks6 to identify these opportunities. The first maps where to look across the value chain of enablers, monetisers, and adopters. The second sets out what to look for in individual companies: a durable moat, management with both vision and the ability to execute, and momentum—early wins that compound into lasting advantage.

Both are getting harder to apply. The boundaries are blurring: chip makers are backing model developers, and model developers are building data centres and selling industry-specific services. A single company can sit in two of our categories at once. Value is also spreading unevenly: the model layer is consolidating around a few players while commoditisation pressure rises elsewhere, and adoption varies widely by industry and geography. So granularity matters more than ever: we judge assets one at a time rather than buy the theme.

Software is a case in point. There was a fear that AI would hollow out the entire sector. Earnings so far have not borne that out for firms with proprietary data, mission-critical workflows, and deep customer integration.

Safety is another. The stakes are not only financial, and I doubt anyone needs persuading of that. How fast to go is being debated among the people building these systems, and they do not agree. As investors, we have to take a view on the whole sector as well as company by company.

Energy: Where strain meets opportunity

My third topic follows directly from the second. I shared earlier this year that energy is becoming one of the most constrained and contested parts of the global economy, due to three competing demands: AI, geopolitics, and the climate transition7. Grid investment lags demand growth, and permitting delays and equipment shortages are slowing the build-out.

That strain is also where much of the investable opportunity sits. Countries are taking divergent paths as they weigh security, affordability, and climate commitments, so energy costs and industrial competitiveness now vary widely from market to market. That is why we look for particular characteristics rather than particular sectors: regulated networks with inflation and volume protection, technologies that ease congestion where the build-out is slow, and equipment businesses with real pricing power. What these have in common is that demand for them does not depend on which energy pathway a country chooses.

On climate more broadly, progress will not be linear. Policy has shifted in many countries, and the path is likely to stay uneven. But the physical realities are accelerating regardless, and it is widely accepted that the world is unlikely to hold warming below 1.5 or 2 degrees. The structural drivers are intact: energy security, decarbonisation, and adaptation.

We sized the adaptation opportunity with Bain last year8. Annual revenues from a selected set of solutions could grow from about US$1tn today to US$4tn by 2050, with the corresponding investment universe rising from US$2tn to US$9tn. Those are 2050 numbers and the market is still nascent, so we approach it the way we approach the climate transition more broadly: investing in established solutions where demand is already real, with a smaller and carefully managed allocation to emerging innovations.

The advantage of patient capital

Those three claims—governments, the AI build-out, and energy systems—are competing for the same pool of long-term money. That is what I meant about patience being repriced.

The principles I keep returning to over the last five years of the Summit have not changed: clarity of purpose, diversification, granularity, agility, and partnerships. The environment is testing all five. The advantage of long-term capital—whether it is a family’s wealth or a country’s reserves—is that we do not have to predict the outcome. We have to be prepared for it, selective in what we own, and ready to move when others cannot.