New York, Oct 3: The global artificial intelligence race is entering a new phase as technology companies, investors and governments pour unprecedented amounts of money into computing infrastructure, data centres and advanced AI systems, even as questions grow over whether the expected economic returns will arrive quickly enough to support the spending.
The scale of investment has become one of the defining features of the technology industry in 2026. Companies developing advanced AI models are expanding their computing capacity, while major cloud providers are committing enormous sums to data centres and specialised hardware needed to train and operate increasingly sophisticated systems.
A Reuters analysis published on October 3 highlighted the extraordinary size of the infrastructure buildout. Global spending on data centres could exceed $30 trillion by 2050, according to a projection cited from PwC. The figure illustrates how deeply artificial intelligence is beginning to influence long-term investment decisions across the global economy.
The rapid expansion has also created expectations that AI will generate a new wave of productivity and economic growth. Technology companies and their investors argue that increasingly capable systems could transform office work, scientific research, software development, manufacturing and other industries.
However, economists and analysts are raising a fundamental question: will the applications and revenues created by AI develop quickly enough to justify the enormous capital being deployed today?
The concern is not about whether artificial intelligence can become an important technology. Few observers doubt its potential. Instead, the debate centres on the timing and scale of its economic impact.
Large technology companies are currently spending heavily on computing infrastructure because demand for AI services continues to rise. Training advanced models requires enormous quantities of computing power, while operating those systems for millions of users creates another substantial requirement for chips, electricity, networking equipment and data-centre capacity.
This has created a powerful investment cycle. More AI users create greater demand for computing capacity. That demand encourages companies to build additional infrastructure, which in turn raises expectations for the future growth of AI applications.
Yet the economic return from those investments remains difficult to measure across the entire economy.
JPMorgan has said broad-based productivity gains in the United States, which currently leads the AI race, remain difficult to identify at the scale required to justify some of the market expectations surrounding the technology, according to the Reuters report.
Bain & Company has similarly argued that productivity improvements from existing markets may not be sufficient to support current levels of AI infrastructure spending. The consulting firm has suggested that entirely new markets may have to emerge to close the gap between investment and potential returns.
Those new markets could include AI-controlled robotics, advanced manufacturing, scientific discovery and technologies capable of creating new materials. Such developments could eventually produce large economic benefits, but their commercialisation may take considerably longer than the infrastructure expansion now taking place.
The issue is particularly important for companies operating large-scale data centres. Hyperscalers and other technology firms must continue spending billions of dollars to keep up with demand while simultaneously generating enough revenue to make those investments financially sustainable.
According to Bain, major US hyperscalers and other participants in the AI infrastructure race may need to generate more than $4.2 trillion in additional revenue over the next five years to fund the expected buildout.
That figure highlights the central challenge facing the sector. Building computing infrastructure is only one part of the AI business model. The systems eventually need to create enough value for businesses and consumers to support the cost of developing and operating them.
For now, enthusiasm surrounding AI remains strong. Investors continue to view artificial intelligence as one of the most important technological shifts of the decade. Global equity funds attracted $34.76 billion during the week through September 30, with Reuters reporting that optimism around AI spending was among the factors supporting investor demand.
Goldman Sachs has estimated that major US hyperscalers could spend about $800 billion on capital expenditure during 2026, with market expectations pointing toward approximately $1.1 trillion in 2027. Strong demand for cloud services and limited infrastructure supply are helping sustain the investment cycle.
The semiconductor industry is also benefiting from the expansion. AI systems require powerful processors and high-bandwidth memory, while data centres need increasingly advanced networking equipment and energy infrastructure.
This demand is creating opportunities well beyond companies that develop AI models. Semiconductor manufacturers, memory producers, cloud operators, fibre-optic suppliers and power infrastructure companies are all becoming part of the expanding AI ecosystem.
At the same time, the rapid growth has raised concerns about concentration of capital and the possibility that some investments may ultimately fail to produce the anticipated returns.
AI companies face a difficult balancing act. They need to spend aggressively to remain competitive because falling behind in computing capacity can weaken their ability to develop and deploy advanced models. But excessive spending without corresponding revenue growth could put pressure on valuations and financing.
The technology sector has experienced similar periods of intense investment before. The internet boom of the late 1990s produced enormous infrastructure spending and unrealistic expectations in some areas, but it also laid the foundation for businesses that later transformed the global economy.
The comparison does not necessarily mean the current AI boom will end in the same way. Instead, it illustrates how technological revolutions can require years before their full economic benefits become visible.
Artificial intelligence may eventually produce major gains in productivity, research and automation. The immediate challenge is determining how much capital is required to reach that future and how long investors must wait before those gains translate into sustainable revenue.
As companies continue building larger data centres and developing increasingly powerful models, the AI industry is therefore moving beyond a simple race for technological capability.
It is becoming a test of whether technological ambition, infrastructure spending and commercial demand can develop at the same pace.
The coming years will determine whether today’s enormous AI investments become the foundation for a new era of economic growth or whether parts of the spending cycle prove to have moved faster than the underlying market.