Investors and policymakers at two major Singapore conferences today broadened their concerns about artificial intelligence beyond stretched valuations, warning that the technology’s rapid development could bring risks ranging from loss of human control to bioterrorism and a potentially destabilising investment bubble.
The shift in tone comes as spending on AI infrastructure reaches unprecedented levels, forcing investors to confront a difficult question: whether rapidly rising revenues and productivity gains will eventually justify the enormous capital being committed to chips, data centres and computing capacity.
“Anyone who tells you I’ve got absolute control over my autonomous agent… obviously doesn’t know what he’s talking about. So loss of control is a real threat,” Singaporean Foreign Minister Vivian Balakrishnan said at the Milken Institute Asia Summit.
Balakrishnan said the possibility of state and non-state actors using AI to develop new weapons was also a growing concern, saying the risk “whether it’s in chemistry or biology or in weapons of mass destruction, is very real.”
Kiran Mazumdar-Shaw, founder and chairperson of Indian biopharmaceutical firm Biocon, said AI could be used by malicious actors to design new viruses or pathogens.
“The real existential threat is bioterrorism,” she said at the Forbes Global CEO Conference, calling for guardrails and governance. “I don’t think self-governance is going to work.”
The warnings come as AI investment accelerates rather than slows. Gartner forecasts global AI spending will reach $2.7 trillion in 2026, up 49.5% from 2025, with AI-optimised servers, data-centre infrastructure and semiconductors accounting for a large share of the increase.
Stanford University’s 2026 AI Index found that global corporate AI investment more than doubled in 2025 to about $581.7 billion, while private investment rose 127.5%. Compute costs for leading AI companies have also climbed sharply as developers build increasingly powerful models.
The scale of infrastructure spending is particularly striking. The combined capital expenditure of eight major cloud-service providers, including Microsoft, Alphabet, Amazon, Meta and Oracle, is expected to exceed $710 billion in 2026, according to TrendForce.
That spending has created a widening gap between the money flowing into AI infrastructure and the still-developing ability of companies to monetise the technology at scale.
Only 22% of organisations surveyed by Gartner said they had successfully scaled AI across multiple business units or adopted an AI-first approach. Yet 85% of functional leaders said they planned to increase AI spending in 2026.
“If the AI narrative is unwound because of safety issues, regulation or because as we go through 2027, end users don’t see sufficient ROI (return on investment), there could be issues,” said Rohit Sipahimalani, chief investment officer at Singapore state investor Temasek.
“The biggest risk to the market is the AI narrative. Right now, it seems very strong,” Sipahimalani said at the Milken Institute event.
The concern is not that AI lacks economic value. Rather, investors are increasingly questioning whether the economic returns will arrive quickly enough to support the valuations and infrastructure commitments built around the technology.
JPMorgan Asset Management estimates that generating a 10% return on current AI investment could eventually require about $650 billion in additional annual revenue. It estimates hyperscalers will spend about $533 billion on capital expenditure in 2026, underscoring the scale of the monetisation challenge.
The technology’s supporters can point to rapidly expanding revenues and rising corporate adoption. Stanford’s AI Index found that leading frontier AI companies are reaching meaningful revenue levels unusually quickly, even as compute and infrastructure costs continue to rise.
But that equation remains particularly difficult for the model developers themselves. Reuters reported this week that Anthropic, one of the leading AI companies, generated $4.6 billion in revenue in 2025 while spending $7.3 billion on compute and infrastructure, contributing to an $8.06 billion operating loss.
Bridgewater Associates founder Ray Dalio was more direct about the investment risks, describing AI as a “classic bubble” and saying there was a difference between “how terrific a technology is and how terrific an investment is.”
He noted that rising rates and changing circumstances had not “yet had a negative effect on other things,” but that there was pressure for interest rates to rise further.
Once higher rates began to bite, the bubble would start to burst, Dalio said. “I think we’re close to that.”
The financing environment is becoming an important part of the debate. S&P Global Ratings said in August that combined capital expenditure by the six largest hyperscalers could exceed $1.3 trillion by 2027, while negative free operating cash flow was expected across the group in 2026 and 2027. Debt, leases, guarantees and other financing structures are also becoming more important in funding the AI buildout.
That does not necessarily mean the AI boom is a conventional bubble. Blackstone said in its 2026 investment outlook that major AI infrastructure spending differs from earlier speculative booms because much of the investment is being funded by the cash flows of large technology companies with strong balance sheets. It estimated the top five hyperscalers spent about $415 billion on capital expenditure in 2025 against combined revenue of about $1.7 trillion.
The greater risk may therefore lie in the speed and concentration of the investment cycle rather than in AI technology itself. If demand for computing capacity continues to grow and AI applications generate sufficient revenue and productivity gains, today’s infrastructure spending could become the foundation of a new technology cycle.
If adoption, monetisation or productivity disappoints, however, the enormous capital commitments could leave investors facing a sharp reassessment of valuations.
Singapore is particularly exposed to that question because of its role as a financial centre and regional technology hub. The government said last month that AI-related activity was expected to contribute significantly to the country’s economic growth, while acknowledging the risks to the economy from a major correction in global AI investment and asset valuations.
The Singapore discussions offer a snapshot of a technology boom entering a more complicated phase. AI is no longer being judged solely on whether it works. Investors are increasingly asking whether it can generate sufficient returns, policymakers are asking how its risks can be contained, and companies are being forced to justify infrastructure spending that runs into hundreds of billions of dollars.
For now, capital continues to flow. The debate has shifted to what happens if the returns do not arrive fast enough.
Stay ahead of the Stories shaping our world. Subscribe to Impact Newswire and join our
WhatsApp Channel for updates on global tech, business, and innovation—all in one place.
Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.
Got a story to share? Contact Us to reach a global audience with Impact Newswire.
Discover more from Impact AI News
Subscribe to get the latest posts sent to your email.
