MAS Warns AI Valuations Echo Dot-Com Bubble Risks

The Monetary Authority of Singapore (MAS) has sounded a clear warning in its annual Financial Stability Review, stating that equity markets are witnessing ‘relatively stretched valuations concentrated in the technology and AI sectors.’ The regulator emphasized that much of the recent rise in global equity markets has been driven by AI-linked investments, creating concentrated exposure for many investors. This assessment marks one of the most explicit regulatory comparisons between current market conditions and the speculative excesses of the late-1990s dot-com bubble.

Particular concern was raised about the financing methods employed by major technology firms. The MAS identified ‘novel and potentially circular private financing arrangements’ including special purpose vehicles, private credit structures, and unconventional accounting treatments that could mask leverage and increase funding dependencies. These opaque structures, primarily used by hyperscalers to fund expansions, represent potential systemic risks that could amplify market vulnerabilities during a downturn.

The scale of the AI valuation surge is staggering. OpenAI, creator of ChatGPT, recently achieved a $500 billion valuation and is reportedly targeting $1 trillion ahead of a potential 2026 IPO. Meanwhile, Anthropic has nearly tripled its value since March, skyrocketing from $60 billion to $170 billion. This breakneck pace of valuation growth has occurred despite what Jordi Alexander, CEO of trading firm Selini Capital, describes as ‘game-altering productivity gains from AI expected still in the distant horizon.’

Alexander explained to Decrypt that the economy requires high growth rates to sustain elevated government debt levels, and with most sectors unable to deliver such growth, the AI sector has absorbed massive investment and attention. ‘With the game-altering productivity gains from AI expected still in the distant horizon, questions of a temporary AI bubble are fair to ask,’ he said. ‘Many major AI companies will be financially exposed if the compounding revenue story for them does not play out.’

Nirav Murthy, co-founder and co-CEO of Camp Network, acknowledged that valuations have outpaced fundamentals but maintained a more nuanced perspective. ‘We’re in a phase where capital intensity, circular deal structures, and opaque accounting can make growth look inevitable when it’s really just well-financed,’ he told Decrypt. However, he cautioned that ‘the next leg has to be earned with real unit economics,’ suggesting that sustainable profitability must eventually validate current valuations.

The MAS warning extends beyond simple valuation concerns to highlight structural weaknesses in the AI ecosystem. Murthy noted that if investor sentiment cools, ‘the pain will show first in long-duration equities and private credit linked to data-center buildouts.’ This specific vulnerability points to potential contagion effects that could spread beyond pure AI companies to infrastructure providers and financiers.

Perhaps the most underappreciated risk, according to industry experts, involves unresolved intellectual property disputes. Murthy warned that ‘we’re seeing models trained on questionable datasets, rights disputes kicked down the road, and legal risk treated as a line item.’ He emphasized that for sustainable profitability, major players must ‘lock in rights-clean, clearly licensed, provenance-verified data as core infrastructure.’ This legal uncertainty represents a significant potential liability that could undermine the entire sector’s financial foundation.

Despite these concerns, Murthy acknowledged that certain segments of the AI stack—particularly chipmakers and major platforms—remain profitable. This differentiation suggests that while the broader AI sector may face valuation pressures, companies with established revenue streams and clear technological advantages might weather any potential downturn better than their more speculative counterparts.