AI Capital Inflows Surge as Productivity Gains Remain Elusive
Global spending on AI infrastructure could reach $30 trillion by 2050, but economists and analysts question if this will yield sufficient returns for investors.
Massive Capital Inflows
Global investment in artificial intelligence (AI) infrastructure is projected to reach $30 trillion by 2050 for data centres alone, according to PwC. This figure, which PwC stated could nearly match the current value of outstanding US Treasuries, dwarfs capital outlays seen during the railway or internet revolutions, even after inflation adjustments.
Separately, AI firm Anthropic plans to invest $518 billion in coming years, as revealed in its IPO prospectus. The same document shows this sum is over 100 times its 2025 revenue.
Elusive Productivity Gains
Despite the vast capital influx, economists question if AI applications will generate sufficient returns. JP Morgan noted in August that broad productivity gains, particularly in the US, remain “elusive”.
A Bain & Company study, published last month, suggested that existing markets alone cannot justify current outlays; it called for over $4.2 trillion in new revenue within five years to fund the build-out by hyperscalers like Google, Amazon, and Microsoft.
Columbia Business School economist Stijn Van Nieuwerburgh estimated US AI investment at $9 trillion from 2025 to 2032, requiring $3.55 trillion in annual revenue by 2032 for a 10 per cent return. This is equivalent to 3.2 per cent of US GDP each year.
High Valuations and Funding Risks
High valuations for AI companies hinge on substantial future productivity increases, which have yet to materialise broadly. JP Morgan estimated that US productivity would need to grow 3 per cent to 5 per cent annually over the next decade to justify Nvidia’s valuation.
This significantly exceeds the US Congressional Budget Office’s baseline expectation of 1.75 per cent annual growth for that period. Google DeepMind's Chief Strategy Officer, Jasjeet Sekhon, stated in August that recursive self-improvement, where AI models enhance themselves, forms a “key part of the investment thesis”.
However, economist Diane Coyle of Cambridge University observed that productivity changes might still lag corporate accounting timelines. Columbia Business School's Van Nieuwerburgh also highlighted risks from leveraged debt funding, where modest demand dips or delays could lead to larger losses.
Asian economies and companies, increasingly central to global technology supply chains and data centre expansion, face similar scrutiny on AI returns. While specific figures for Asia are not detailed, the global trend of high investment and elusive productivity gains suggests that Asian investors and technology firms should closely evaluate their AI strategies.
The pressure on US hyperscalers to find over $4.2 trillion in new revenue in five years will likely drive intense competition for AI application markets globally, including across Southeast Asia.
This may accelerate the need for Asian businesses to develop novel AI-driven applications that demonstrably generate new revenue streams and justify the substantial infrastructure outlays.
This article is journalism, not investment advice; consult a licensed professional before making financial decisions. Market data is indicative, may be delayed, and should be verified with your broker or exchange before use.
Comments.
Comments are moderated. We remove what is unlawful, abusive or off-topic, and and you remain responsible for what you post.
Reader comments open soon. Until then, corrections and responses go to our newsroom, and we publish what we get wrong on Corrections.