Reuters Projects $30 Trillion AI Spend by 2050 on Risky Productivity Assumptions

Profits for the next generation of computing depend on whether the industry can translate massive capital outlays into actual productivity. Total spending on artificial intelligence and data centers may reach $30 trillion by 2050, based on reports from Reuters, though economists warn that returns rely on volatile assumptions. JPMorgan estimates that maintaining Nvidia's current valuation requires U.S. productivity to grow by 3% to 5% annually over a decade, far above the Congressional Budget Office baseline of 1.75%. Individual firms face similar strain, as Anthropic has $518 billion in future infrastructure obligations, more than 100 times its 2025 revenue of $4.6 billion.

These liabilities underscore a broader industry revenue gap identified by Bain & Co., which finds that AI must generate $6 trillion in annual revenue by 2031 to justify its data center expansion. Current enterprise and consumer services are projected to provide only $1.2 trillion to $1.8 trillion, leaving a shortfall of roughly $4.2 trillion that must be bridged by new markets in robotics, energy generation, and drug discovery. Bain estimates that funding the sector sustainably would require adding approximatel