Major technology firms including Microsoft, Alphabet and Meta Platforms are set to release spending updates this week that highlight heavy investment in AI infrastructure—from data centres to specialised chips. According to Barclays strategists, the annual capex of just these five hyperscalers doubled from 2022 to 2024 to over USD 200 billion, with projections suggesting a rise to about USD 500 billion by 2027.
Investors are treating capex as a barometer of belief in the AI theme: sustained spending signals commitment, while pull-back or cautious commentary could prompt a re-rating of the broader tech rally. Amid this, the market milestone achieved by Nvidia—surpassing a USD 5 trillion market capitalisation—amplifies scrutiny on whether large-scale AI capex is justified by future returns.
However, analysts caution that the massive infrastructure outlays carry risk. Past cycles saw large build-outs ahead of demand and oversupply, a theme re-mentioned by portfolio managers drawing parallels with the late-1990s dot-com bubble.

Key Highlights
• Tech giants’ AI-related capital spending is being closely watched as a signal of long-term conviction.
• Annual capex among major hyperscalers rose from ~USD 200 billion in 2024 to a projected ~USD 500 billion by 2027.
• Nvidia’s rise to USD 5 trillion market value underscores the centrality of AI hardware and infrastructure in the current market narrative.
• Investors warn that while the infrastructure demand is real, returns may lag, increasing risk of over-investment.

Why This Matters
• Heavy infrastructure spending suggests that the next phase of the AI cycle is deeply tied to hardware, data-centres and global build-out, not just software innovation.
• If companies scale back or signal caution in their capex commentary, it may trigger a broader tech-market rotation away from infrastructure-heavy names.
• Large, sustained investment in data-centres and AI hardware could exacerbate supply-chain bottlenecks, pricing pressures and international competition for resources.
• For markets, this dynamic raises the prospect that current valuations of the ‘AI theme’ may be vulnerable if the expected monetisation of infrastructure doesn’t match upfront investment.

Source
Reuters – Full Article