Temasek Assets Surpass $400 Billion as AI Exposure Doubles in Five Years
Taylor Wilson
Temasek's net portfolio hit a record $401 billion, delivering a 10.5% annual return; the bigger signal is its plan to lift AI exposure from 6% to 15% by 2031 — one of the world's largest sovereign funds is making AI its defining bet for the next decade.
How big is $401 billion?
As of March 2026, Temasek's net portfolio reached S$518 billion (≈US$401 billion), a record high for the second consecutive year, placing it among the world's top ten sovereign wealth funds.
Full-year total shareholder return was 10.5% in Singapore-dollar terms, or roughly 16.4% in US-dollar terms — nearly matching the S&P 500 over the same period.
This means → an Asia-headquartered sovereign fund matched the US equity benchmark in dollar terms, while Canada's CPPIB posted 7.8% and British Columbia Investment Management 6.7%.
AI exposure from 6% to 15% — where does the money go?
Temasek plans to raise AI-related exposure from 6% to 15% of its portfolio by 2031 — more than doubling. In plain terms = out of every $100 invested, the AI slice rises from $6 to $15.
Five focus areas: energy and data centres, semiconductors, cloud providers, foundation models (OpenAI / Anthropic), and AI applications and software infrastructure. Existing holdings include Nvidia, Microsoft, TSMC, Tencent, and Amazon.
This year's action: Temasek joined OpenAI's $122 billion funding round and Anthropic's $65 billion round. This reflects real capital flowing into foundation-model companies, not just strategic rhetoric.
Is Nvidia's valuation a bubble?
CIO Rohit Sipahimalani said publicly that Nvidia's P/E is lower than many popular tech stocks. "Is this a bubble? I don't think so. But maybe some other companies are, so you have to look at each one."
His risk-management logic: holding listed AI stocks gives liquidity — if valuations stretch too far, Temasek can trim positions flexibly and redeploy capital elsewhere.
This means → Temasek's AI build-up strategy is "listed equities first, private equity second." Per the Financial Times, new AI investments will come mainly through public-market purchases; as AI's share rises, other tech and telecom holdings will shrink proportionally.
Americas double in a decade, China pivots to hard tech — how is the map shifting?
Americas exposure has risen to 26% of the portfolio, up from roughly 11% in 2016, making it the second-largest market after Singapore (27%). Sipahimalani said current priority opportunities are "global, but tilted slightly toward the US."
China exposure grew by S$10 billion (≈US$7.7 billion) this fiscal year — the largest single-year increase in five years — lifting its share modestly to 17%, still well below the 24% level a decade ago.
In plain terms = Temasek has not exited China, but the direction of capital has changed — away from consumer and real estate, toward AI hardware, robotics, biotech, and energy transition. New investments include Luckin Coffee and logistics group ANE Logistics.
Private credit and infrastructure — what are the other two 5% targets?
Private credit exposure is set to rise from 2% to 5% by 2031, focused on senior secured structures covering corporate loans, asset-backed finance, and real-estate credit. In plain terms = lending money to companies or projects for interest income, but requiring collateral as a backstop.
On infrastructure, the target for "core-plus" assets — renewables, nuclear, energy storage, decarbonisation tech — moves from roughly 1% to 5%.
Head of global investments Nagi Hamiyeh outlined the approach: fewer but larger bets, with each private deal typically no less than $200 million, requiring a significant minority stake and deeper strategic involvement.
Three lines advancing at once — what is the core risk?
Temasek's capital allocation over the next five years tilts simultaneously toward three themes: AI infrastructure, China hard tech, and alternative credit. This reflects a shift from passive, market-following allocation toward active, conviction-driven investing.
This fiscal year, new investments totalled S$39 billion against S$31 billion in divestments (including a S$8.18 billion sale of Schneider Electric India), leaving a net positive deployment.
This means → whether the entire strategy delivers depends on two variables: whether AI commercialisation keeps pace with expectations, and whether geopolitical friction disrupts the cross-border capital flows on which the plan relies — the question markets will keep watching.
Content is for reference only, not financial advice.