Why Temasek Is Comfortable With Hyperscalers’ AI Spending

Temasek’s confidence in hyperscalers’ AI spending rests on a bet that adoption will grow enough to generate risk-adjusted returns—and on the financial strength of the companies funding the build-out. The Singapore investor plans to increase its AI-related portfolio exposure from about 6% to as much as 15% by 2031. That is a strategic target, not evidence that every AI infrastructure investment will pay off.

Why Temasek is comfortable with the spending

Temasek North America head Jane Atherton said the investor was comfortable with current hyperscaler spending. Her reasoning, as reported by Reuters, was that the companies underpinning the build-out have strong balance sheets and expect their investments to produce risk-adjusted returns as AI adoption expands.

That makes adoption central to the investment case: infrastructure spending is easier to justify if demand for AI services grows and companies can earn returns on the capital they deploy. But confidence in that outcome is not a guarantee. Temasek’s position is a view about the potential returns, not proof that the spending will be profitable.

Bond-market concerns remain a separate risk

The scale of AI infrastructure investment has also raised concerns about financing. Market commentary has warned that hyperscalers’ borrowing for the build-out could add to bond supply and contribute to fixed-income market volatility.

Those concerns do not, by themselves, contradict Temasek’s view of the companies’ long-term prospects. But they do underline why strong balance sheets and risk-adjusted returns matter: capital-intensive projects still face financing and execution risks. The available reporting does not show that Temasek considers higher interest rates or bond-market pressure irrelevant; rather, its stated comfort focuses on the companies’ financial strength and expected returns.

What Temasek’s AI target means

Temasek says AI-related investments currently represent about 6% of its portfolio and sets a target of up to 15% by 2031. It plans to invest across the AI value chain, including energy and data centres, semiconductors, cloud service providers, foundation models, and AI applications and software infrastructure.

The investor already has exposure to US-based AI companies including Anthropic and OpenAI. The broader plan is not limited to model developers: it includes the infrastructure and services needed to build and deploy AI as well. 8

Portfolio resilience is not the same as AI safety

Temasek has described generative AI as bringing both opportunities and risks, and says it is working to manage its portfolio companies’ exposure to AI-related change. That speaks to investment and business resilience; the available statements do not establish a specific position on technical AI safety standards. 14

Separately, Temasek’s chief investment officer said the firm would keep 25% of its portfolio in liquid assets to weather shocks and respond to changing opportunities, including those related to AI. That is a financial buffer, not a guarantee against losses from AI investments. 3

How the AI push fits Temasek’s Americas footprint

Temasek’s AI strategy sits within a globally deployed portfolio. Its 2026 speaker notes say almost three-quarters of its underlying exposure is outside Singapore. A report on the portfolio’s regional mix put the Americas at 26%, close to Singapore’s 27%, and identified the Americas as a major investment market.

That international footprint helps explain why US-based AI firms are part of the current portfolio, while the planned increase spans companies and infrastructure across the wider AI value chain. The central test remains whether AI adoption can support returns commensurate with the capital and risks involved.