AI Data Centre Gold Rush Three Trillion Pounds in Play but Bubble Fears Linger

The artificial intelligence (AI) revolution has ignited an extraordinary investment surge, with global data centre spending projected to reach an astonishing $3 trillion by 2028. These vast warehouses, humming with servers, form the core infrastructure powering AI advancements like OpenAI’s ChatGPT and Google’s Veo 3. For now, optimism prevails across financial markets and in the boardrooms of technology giants, where records are being broken. Silicon Valley chipmaker Nvidia recently became the world’s first $5 trillion company, while Microsoft and Apple surged past $4 trillion valuations. OpenAI, following a restructuring, now carries a valuation of $500 billion, hinting at a possible $1 trillion public listing within a year.

The resonance of the AI boom reaches far beyond Silicon Valley. In Newport, Wales, Microsoft’s new data centre on the site of a former radiator factory symbolises this economic shift. Local leaders and residents view these projects as a bridge to future prosperity, a hopeful answer for communities seeking to replace traditional industries with high-tech opportunity.

However, questions are building around the sustainability of such prolific investment. Amazon, Meta, Google and Microsoft—the so-called hyperscalers—are pouring more than $750 billion into AI-related capital expenditure over the next two years alone. The magnitude has been described by US investment firm Manning & Napier as “nothing short of incredible”. Yet beyond these tech giants, smaller and more speculative players are increasingly reliant on private credit. Analysis from Morgan Stanley estimates that $1.5 trillion of the projected global spend will need to come from non traditional sources, potentially introducing new layers of systemic risk should debt-backed projects falter.

Not all industry voices are uniformly confident. Joe Tsai, chair of Alibaba, has already warned of excesses in the data centre market, citing construction projects launched without secured commitments from future customers. With 11,000 data centres now operating globally—a 500 percent rise in two decades—fears are surfacing of overbuilding or empty promises. The Uptime Institute, a leading global inspector of these facilities, believes many announced sites will remain partially filled or may never be constructed at all.

Behind the outlays lie ambitious revenue expectations. Morgan Stanley anticipates generative AI revenues—from chatbots to AI agents and image generators—to rocket from last year’s $45 billion to $1 trillion by 2028, contingent upon robust demand from both the public and private sectors. Yet, research from MIT recently suggested that 95 percent of organisations piloting generative AI saw no return on their investments, prompting market jitters over the medium-term business adoption of these technologies.

Despite uncertainties, data centres are by no means AI-exclusive. As Microsoft’s Alistair Speirs points out, these facilities underpin a wide range of digital services, from cloud storage and email to routine video conferencing, demonstrating their increasing role as a general-purpose infrastructure for the modern economy. Major projects dedicated solely to AI, including the $500 billion Stargate initiative in Texas as well as new ventures in the UK and US, highlight the scale—and risk—of current commitments.

With global data centre capacity set to double by 2030, the expansion is driving parallel investments in power infrastructure, forecast by Goldman Sachs to absorb an additional $720 billion. For regions like Newport, the influx of technology and talent is reshaping local economies. As the world bets billions on the ongoing AI revolution, a careful balance between innovation and prudence will be crucial to ensure that the boom does not give way to a bust.