‘Phenomenal’ amount of capital going into AI could be more spread out: Temasek CEO

Temasek's Dilhan Pillay Sandrasegaran and DBS' Tan Su Shan spoke at the Forbes Global CEO Conference, where American billionaire investor Ray Dalio also warned of a "classic bubble".

Temasek Holdings chief executive officer Dilhan Pillay Sandrasegara during the Forbes Global CEO conference at Shangri-La, on Oct 7, 2026. (Photo: CNA/Ili Mansor)

This audio is generated by an AI tool.

SINGAPORE: With a “phenomenal” amount of money going into artificial intelligence, the capital could be spread more widely to areas such as AI adoption, Temasek Holdings CEO Dilhan Pillay Sandrasegaran said on Wednesday (Oct 7).

Speaking on a panel at the Forbes Global CEO Conference, the head of Singapore's sovereign wealth fund said much of the money is flowing into chips and infrastructure. The panel included billionaire Ray Dalio, Franklin Templeton CEO Jenny Johnson and Bangkok Bank president Chartsiri Sophonpanich.

Mr Sandrasegaran questioned whether the cost of capital will rise because of inflationary pressures from investment in AI infrastructure.

“So much (is) moving into AI infrastructure and so on that we have to start thinking now about how much capital is available for the adoption going forward,” he said.

Temasek is considering this now because its companies need to borrow for their operations, transformation and AI adoption, he said.

Capital is flowing into the five layers laid out by Nvidia CEO Jensen Huang: energy, chips, infrastructure, models and applications, said Mr Sandrasegaran. Much of the profit pool has gone to the chip sector.

“The memory chip guys are making a bundle now compared to any time in the history of semiconductors,” he said.

He also addressed the decision by ST Telemedia, a Temasek subsidiary, to sell its 82 per cent stake in ST Telemedia Global Data Centres to investment firm KKR and Singtel for S$6.6 billion (US$5.16 billion).

“People asked me, ‘Why would you want to sell a data centre operator when all that money is going to AI right now?’” said Mr Sandrasegaran.

The subsidiary relied on Temasek's triple-A rating to get good credit at good rates, he said, adding that costs would have risen had it continued to invest in energy capacity or better technology for its data centres.

“You have to think about the morning after when you're the owner of a business, and then your risk tolerance level is different from that of an investor, because the ability to rotate is not the same,” he added.

EXPERIMENTATION AND TRUST IN AI ADOPTION

On a later panel, DBS Group CEO Tan Su Shan discussed how companies can encourage AI adoption.

The panel, which covered the growth of technology amid geopolitical and workforce changes, included Northeastern University president Joseph E Aoun, Singapore Exchange chairman Koh Boon Hwee, Trip.com Group CEO Jane Sun and SM Investments vice-chairperson Teresita Sy-Coson.

One way to drive AI adoption is to allow experimentation and failure in a safe environment, said Ms Tan. DBS gives employees a safe space grounded in enterprise knowledge, with controls and governance policies, to build their own AI agents, she added.

CNA Games Guess Word Crack the word, one row at a time Buzzword Create words using the given letters Mini Sudoku Tiny puzzle, mighty brain teaser Mini Crossword Small grid, big challenge Word Search Spot as many words as you can Show More Show Less

DBS Group chief executive officer Tan Su Shan during the Forbes Global CEO conference at Shangri-La on Oct 7, 2026. (Photo: CNA/Ili Mansor)

When she took the helm of the bank, she challenged teams to come up with the best AI agent others could use, and units to create “big, massive operating model transformation changes”.

Ms Tan also highlighted the importance of data governance and accountability, given the large volume of existing legacy data and new data created every day. Ownership of the data lifecycle is often unclear, and the way data is tagged, stored and used will have to change, she said.

“So first of all, it's really creating a foundation of making sure that your end-to-end data, from the day it's born to the day it's purged, you have clear ownership, accountability and you know where it sits. And then a rubric around how you deal with structured data and unstructured data,” she said.

A firm's culture is another factor, said Ms Tan. It must stay focused on doing the right thing for the customer rather than just optimising profits.

Protecting people in the face of job transformations must also be considered, she said.

“For us, we have a rhetoric that says we're here not to protect jobs, we're here to protect people,” she said, adding that employees need the right attitude to harness AI and move up the productivity ladder.

For instance, someone in operations whose role disappears can look at becoming an AI agent manager – the human in the loop who watches the AI agents as they operate, she said.

DALIO WARNS OF 'CLASSIC BUBBLE'

Mr Dalio told the audience that AI is a "classic bubble" close to bursting. The market is at a stage of the cycle where rising interest rates and changing circumstances have yet to hurt other assets, he said.

The bubble starts to pop when rates keep climbing, said Mr Dalio, who has long warned of an AI bubble.

His comments come as PWC analysts predicted that global data centre spending will reach US$31.6 trillion by 2050 to meet the growing appetite for AI. That is more than the United States’ gross domestic product of US$30 trillion.

Overinvestment often accompanies big technological advancements, said Franklin Templeton's Ms Johnson.

“What we’re talking about with the bubble is in these hyperscalers who are deploying a huge amount of capital, raising a ton of debt,” she said.

She stressed that all projections of the energy these investments will need are based on today's technology. Long-term commitments built on those projections are therefore "a little bit more hazardous" amid rapid innovation, she said.

Our chief editor shares analysis and picks of the week's biggest news every Saturday.

Get our pick of top stories and thought-provoking articles in your inbox

Stay updated with notifications for breaking news and our best stories

Join our channel for the top reads for the day on your preferred chat app

淡马锡的迪尔汉·皮莱·桑德拉塞加兰和星展银行的陈素珊在福布斯全球CEO大会上发表了讲话，美国亿万富翁投资者雷·达里奥也在会上警告称存在“典型的泡沫”。

淡马锡控股首席执行官迪尔汉·皮莱·桑德拉塞加拉 (Dilhan Pillay Sandrasegara) 于 2026 年 10 月 7 日在香格里拉酒店出席福布斯全球 CEO 大会。（图片：CNA/Ili Mansor）

这段音频由人工智能工具生成。

新加坡：淡马锡控股首席执行官迪尔汉·皮莱·桑德拉塞加兰周三（10 月 7 日）表示，随着大量资金涌入人工智能领域，这些资金可以更广泛地分散到人工智能应用等领域。

在福布斯全球CEO大会的一个小组讨论会上，新加坡主权财富基金负责人表示，大量资金正流入芯片和基础设施领域。与会小组成员包括亿万富翁雷·达里奥、富兰克林邓普顿首席执行官珍妮·约翰逊和曼谷银行行长查特西里·索蓬帕尼奇。

桑德拉塞加兰先生质疑，由于人工智能基础设施投资带来的通胀压力，资本成本是否会上升。

他说：“很多东西都在向人工智能基础设施等领域转移，所以我们现在就必须开始考虑未来有多少资金可用于人工智能的普及应用。”

他表示，淡马锡现在正在考虑这个问题，因为其旗下公司需要借贷来支持运营、转型和人工智能应用。

桑德拉塞加兰先生表示，资金正流入英伟达首席执行官黄仁勋提出的五个层面：能源、芯片、基础设施、模型和应用。大部分利润都流向了芯片领域。

他说：“与半导体历史上任何时期相比，存储芯片厂商现在赚得盆满钵满。”

他还谈到了淡马锡旗下子公司 ST Telemedia 决定将其在 ST Telemedia Global Data Centres 的 82% 股份以 66 亿新元（51.6 亿美元）的价格出售给投资公司 KKR 和新加坡电信。

桑德拉塞加兰先生说：“有人问我，‘既然现在所有的钱都流向了人工智能领域，你为什么要出售一家数据中心运营商呢？’”

他表示，该子公司依靠淡马锡的AAA评级以优惠利率获得良好的信贷，并补充说，如果继续投资能源容量或数据中心的更好技术，成本将会上升。

“当你成为企业主时，你必须考虑第二天早上会发生什么，那时你的风险承受能力就与投资者不同，因为你的轮动能力是不一样的，”他补充道。

人工智能应用中的实验与信任

在随后的小组讨论中，星展集团首席执行官陈素珊探讨了企业如何鼓励人工智能的应用。

小组讨论的主题是地缘政治和劳动力变化背景下的技术发展，成员包括东北大学校长 Joseph E Aoun、新加坡交易所主席 Koh Boon Hwee、携程集团首席执行官 Jane Sun 和 SM 投资公司副主席 Teresita Sy-Coson。

谭女士表示，推动人工智能普及的一种方法是在安全的环境中允许进行实验和试错。她补充说，星展银行为员工提供了一个基于企业知识、拥有控制和治理政策的安全空间，让他们能够构建自己的人工智能代理。

CNA游戏 猜词游戏 逐行破解单词 流行词游戏 用给定的字母组成单词 迷你数独 小小谜题，挑战你的大脑 迷你填字游戏 小方格，大挑战 单词搜索 尽可能多地找出单词 显示更多 显示更少

星展集团首席执行官陈素珊于2026年10月7日在香格里拉酒店出席福布斯全球CEO大会。（图片：CNA/Ili Mansor）

当她掌管这家银行时，她要求各个团队开发出其他人可以使用的最佳人工智能代理，并要求各个部门进行“大规模的运营模式转型变革”。

谭女士还强调了数据治理和问责的重要性，因为现有的历史数据量庞大，而且每天都会产生新的数据。她指出，数据生命周期的所有权往往不明确，数据的标记、存储和使用方式都必须改变。

“所以首先，要真正建立一个基础，确保你的端到端数据，从创建之日起到删除之日止，都有明确的所有权、责任归属，并且你知道数据存放在哪里。然后，还要制定一套处理结构化数据和非结构化数据的准则，”她说。

谭女士表示，企业文化是另一个重要因素。企业必须始终以客户利益为先，而不仅仅是追求利润最大化。

她表示，在工作岗位发生变化时，保护人们的权益也必须考虑在内。

“对我们来说，我们的理念是，我们来这里不是为了保护工作岗位，而是为了保护人民，”她说道，并补充说，员工需要正确的态度才能驾驭人工智能，并在生产力阶梯上不断进步。

她举例说，运营部门中某个角色消失的人可以考虑成为人工智能代理管理员——也就是在人工智能代理运行过程中监督其运行的人。

达利欧警告“经典泡沫”

达利欧先生告诉听众，人工智能是一个即将破裂的“典型泡沫”。他表示，目前市场正处于周期性阶段，利率上升和环境变化尚未对其他资产造成冲击。

达利欧先生长期以来一直警告人工智能泡沫的存在，他表示，当利率持续攀升时，泡沫就会开始破裂。

在他发表这番言论之际，普华永道分析师预测，到2050年，全球数据中心支出将达到31.6万亿美元，以满足日益增长的人工智能需求。这一数字超过了美国30万亿美元的国内生产总值。

富兰克林邓普顿的约翰逊女士表示，重大的技术进步往往伴随着过度投资。

她说：“我们所说的泡沫，指的是这些超大规模企业投入巨额资金，筹集大量债务。”

她强调，所有关于这些投资所需能源的预测都是基于当今的技术水平。因此，在技术快速创新的背景下，基于这些预测而做出的长期承诺“风险更大”。

我们的主编每周六都会分享对本周重大新闻的分析和精选。

订阅我们的邮件，即可获取精选热点新闻和引人深思的文章。

订阅通知，第一时间获取突发新闻和精彩报道。

加入我们的频道，即可在您常用的聊天应用上获取当日热门文章。