AI Race Heats Up as Global Tech Investment Surges (2026)

Key Takeaways

• Ai race intensifies as the five largest hyperscalers – Amazon, Microsoft, Alphabet, Meta, and Oracle – are projected to spend $660–725 billion on AI infrastructure in 2026, with total spending nearly doubling year-over-year.
• Global AI data centre spending hit $300 billion in 2025, up 60% year over year, with hyperscalers accounting for roughly $216 billion of that amount
• JPMorgan estimates global capital expenditures for AI will total $5.5 trillion through 2030, up from an earlier estimate of $5.1 trillion
• Spending appears to be paying off in the form of sales: global AI sales (excluding China) hit $25 billion in the first quarter of 2026, exceeding the industry’s roughly $21 billion in depreciation costs for the second consecutive quarter
• Power constraints, not capital, are proving to be the new limiting factor as Microsoft alone has an $80 billion Azure order backlog it cannot fill due to power shortages

The Scale of the Investment Surge

Investment in AI infrastructure in 2026 is simply astonishing. If one were to include the 14 largest public data centre operators, capital expenditures (capex) would total $750 billion for the year — more than the GDP of all but the world’s twenty largest economies. Looking further out, the sector is projected to need roughly $3 trillion by 2030 as capacity is doubled. What makes this spending spree remarkable is that, unlike previous technology investment booms, it is much more concentrated: 75% of hyperscaler capex or $450 billion will be spent on AI-specific infrastructure such as GPUs, chips, and data centres.

Who’s Winning the AI Race

Spending is concentrated among a few large firms:

• Microsoft – the company’s recent quarterly spending hit $120 billion or more in fiscal 2026, having already spent $37.5 billion and it has an $80 billion Azure order backlog due to power shortages
• Alphabet/Google – CEO Sundar Pichai was recently concerned about the scale of spending, as the company’s cloud backlog jumped 55% to over $240 billion and it slashed Gemini serving costs by 78% due to model optimisation
• Meta – the firm’s capital expenditures for 2026 are expected to total between $115 billion and $135 billion, including a 1 GW data centre in Ohio and a Louisiana facility that could eventually reach 5 GW
• Amazon and Oracle – both large hyperscalers that participate in the $660–725 billion infrastructure spending boom this year
• Nvidia – set to see over $180 billion in AI chip revenues in 2025 and 2026 combined, making it the largest beneficiary of the AI infrastructure spending boom

Countries Driving the Most Investment

The countries that are set to see the largest investments are also heavily skewed towards the US, as it is where most of the capital expenditures for the AI race are currently allocated. The ways in which the companies balance the massive power demands of AI infrastructure vary: American firms are investing in renewable to co-locate with data centres, Gulf states are using natural gas to fuel new AI infrastructure, and China is building coal, nuclear, and hydroelectric power plants to support its AI ambitions. In other words, the AI race has quickly evolved into an energy and industrial policy competition.

Where is the Money Being Spent?

Most of the money is being allocated to data centres, including real estate, AI-specific chips, and power infrastructure. New data centre construction spending in the US alone hit $45.1 billion per month on average in the last quarter of 2025, up 85% compared to two years ago. At the component level, supply chains for power management chips to regulate power supplies to GPU clusters are expected to remain tight through 2026. The power demands of AI are also extraordinary: training a large language model can consume over 1,000 MWh of electricity, and AI is set to account for 4% of global electricity consumption in 2026.

Corporate Financing & Partnerships

Beyond direct spending, corporate financing arrangements are also changing to meet the needs of the AI infrastructure build-out in the ongoing AI race. JPMorgan estimates that financing for AI-related investments will total $4.1 trillion as companies increasingly turn to loans rather than pure equity to fund infrastructure, and real estate analysts estimate the data centre sector alone will need roughly $870 billion in new debt financing by 2030. Retail investors can also participate in the AI revolution by buying exchange-traded funds, real estate investment trusts, and mutual funds related to the data centre industry.

Risks and Concerns

Some people worry that the rate of investment in AI infrastructure is too high. For example, the payback period for AI infrastructure can be as long as seven years, and although first-quarter 2026 sales figures suggest that revenues from AI infrastructure are beginning to exceed depreciation costs, profit margins are still relatively low. Meanwhile, power shortages are threatening to constrain further investment in AI infrastructure, with major cloud providers such as Microsoft reporting order backlogs due to insufficient power supply. Additionally, regulatory responses to the environmental impact of AI infrastructure could also impact the economics of the AI infrastructure investment boom.

What This Means For the Future

By 2030, AI workloads are set to comprise approximately 50% of total data centre activity, compared to just 25% in 2025. This represents a significant shift, with a large number of data centre operators expecting the transition from training to inference to occur around 2027. Regardless of which firms ultimately dominate the AI landscape, the infrastructure investments in semiconductors, data centres, and power infrastructure will shape the next decade of global technology competition, influencing not only the winners of the AI race but also the broader technology industry.

A Note on Sources

Spending on AI infrastructure is a dynamic area that is constantly evolving as firms update their capital expenditure forecasts each quarter. The figures mentioned in this article are based on a combination of company reports, market research publications such as Intersect360 Research and JLL, and financial reports from Bloomberg, Fortune, and Reuters. All figures are current as of mid-2026. For more analysis on global technology and economic trends, be sure to follow Nexus of Nation.