Spending on artificial intelligence continues to rise, according to the latest forecasts from Gartner, which predicts an increase of 49.7% in 2026. Meanwhile, traditional data center IT spending has been declining since 2023, and this trend could continue.
- AI spending keeps rising, with Gartner forecasting a 49.7% increase in 2026.
- Traditional IT spending has declined since 2023, and AI spending could overtake it in 2026.
- Infrastructure drives growth, especially specialized processors built by hyperscalers.
- Tech giants would need to devote almost all their revenue to AI, so they turn to debt and circular financing.
Growth That Eclipses Traditional IT
According to Gartner’s figures, global AI spending multiplied by 2.6 in 2025, reaching $1.79 trillion. It is expected to rise to $2.67 trillion in 2026, then to $3.64 trillion in 2027, a 36% increase. Over the same period, traditional IT spending fell by 12.5% in 2025, to $3.79 trillion, and is expected to decline by a further 2.5% in 2026 and 9.5% in 2027.
By combining these forecasts with its own estimates of total IT spending, the column’s author believes 2026 could be the first year where AI spending surpasses that of traditional IT. However, the author excludes smartphones and PCs with AI features from the figures, as these capabilities are, in his view, incidental to the purchase of these devices.
Infrastructure at the Heart of Growth
Gartner has slightly raised its forecasts for AI infrastructure, i.e., servers, storage, and switching. Spending for 2025 was revised from $975.6 billion in the May forecast to $981.9 billion in the September one. Gartner adds $52.9 billion for 2026 and $87.4 billion for 2027.
According to the author, this increase is largely explained by specialized processors, or XPUs, designed by hyperscalers—the large cloud service providers—which they use internally and also sell to model developers like OpenAI and Anthropic. John-David Lovelock, Distinguished VP and Chief Forecaster at Gartner, states in the press release accompanying these forecasts: ‘Building data center capacity dedicated to AI is the largest infrastructure project humanity has ever undertaken.’
Revenue to Finance Investments
To assess whether these investments can be financed, the author relies on the work of Synergy Research Group. Its chief analyst, John Dinsdale, anticipates faster revenue growth for the major players from 2026 onwards compared to the 2020-2025 period. Google, Microsoft, Amazon Web Services, Meta Platforms, and ByteDance are among these players. Added together for 2025, 2026, and 2027, their revenues, read from a chart, would reach $6.2 trillion, in IaaS, PaaS, SaaS, search, and social networking services.
These revenues must also cover operating costs, including salaries. Assuming that 75% of AI spending is done by hyperscalers, cloud builders, and model developers, this amounts to $6.1 trillion, according to the author’s calculation, who estimates these players would need to devote almost all their revenue to AI. This is why, he argues, tech giants are resorting to debt, financing from their suppliers like Nvidia, AMD, and Broadcom, or circular operations. The Synergy Research chart indicates that revenues could more than double between 2028 and 2031.
