Tech Firms to Spend $700B on AI Data Centers
Tech companies are projected to spend an unprecedented $700 billion on AI data centers in 2026. NVIDIA CEO Jensen Huang argues that the sector has not yet reached its peak, pointing to continued demand for model training, inference, and edge deployments. This trend is fueling a global infrastructure boom and creating both opportunities and supply chain challenges.
The spending surge is not a short-term blip; forecasts suggest the global data center infrastructure market could exceed $1 trillion in yearly spending by 2030. To meet the demand for AI alone, companies across the value chain may need to invest a cumulative $5.2 trillion to $6.7 trillion by that year. This rapid expansion is creating significant bottlenecks, with power availability being the most prominent obstacle to delivering projects on time. A recent survey found that 83% of industry experts believe local supply chains are not prepared to support the advanced cooling techniques required for high-density AI data centers. These specialized facilities come with a 7-10% construction cost premium over traditional data centers. Beyond chipmakers, a diverse array of companies is reaping the benefits. Infrastructure construction firms like MasTec and Sterling Infrastructure are seeing a boom in business. Similarly, companies specializing in power management and cooling solutions, such as Vertiv, Eaton, and Johnson Controls, are critical to the build-out. The geographic landscape of data center construction is shifting. While established hubs like Northern Virginia remain dominant, developers are increasingly moving to secondary markets in the Midwest and Southeast. This is driven by the search for lower land costs, scalable power, and an effort to avoid community pushback in more densely populated areas. The hyperscale data centers built by Amazon, Microsoft, and Google are massive, with campuses spanning 50-200+ acres. Meta's "Prometheus" project in Ohio, its first gigawatt-scale data center, is expected to come online in 2027 and will feature a 200 MW on-site natural gas power generation facility. This boom is not without financial risks. Some analysts worry about massive overspending on unproven long-term returns. OpenAI alone is projected to burn through $143 billion between 2024 and 2029, a scale of loss unprecedented for a startup.
