Microsoft Lifts 2026 AI Capital Expenditure to $190 Billion
Surging component costs and hardware demand drive $25 billion budget increase
Microsoft has significantly increased its capital expenditure forecast for 2026 to $190 billion, as the company races to secure the hardware necessary for its AI expansion. The revision highlights the escalating physical costs of maintaining AI leadership amidst a global surge in component pricing.
Key details
In its Q3 2026 earnings report, Microsoft revealed that $25 billion of its projected $190 billion annual spend is attributed directly to rising component costs. Prices for critical data center components, particularly memory and storage, have skyrocketed since late 2025—with some hardware more than tripling in price due to intense demand for AI infrastructure.
The company's spending velocity is accelerating rapidly. After investing $32 billion in the last quarter alone to bring more compute capacity online, Microsoft CFO Amy Hood announced plans to spend roughly $40 billion in the upcoming quarter. Despite these massive investments, the company expects to remain supply-constrained through the end of 2026 as demand continues to outpace available hardware.
Why this matters
This massive budget increase demonstrates that the "AI tax" is not just about electricity and water, but also the physical supply chain of compute. When hardware costs triple, the economic threshold for profitable AI services shifts, forcing hyperscalers to commit unprecedented capital just to maintain their competitive position.
Context
The move follows a broader trend of infrastructure bottlenecks across the AI sector. Over the past four quarters, Microsoft has spent approximately $97 billion on infrastructure and equipment to secure $37 billion in annual recurring revenue (ARR) for its AI services. This growing gap between capital investment and immediate returns has led to shifts in pricing strategies, including Microsoft’s recent move to transition GitHub Copilot from flat-rate subscriptions to metered, usage-based billing.
What happens next
Microsoft expects its infrastructure constraints to persist through at least 2026, suggesting that the pressure on the global semiconductor supply chain will remain at peak levels. Investors will be closely watching whether the $190 billion investment begins to close the ROI gap as the company's "pay-per-token" model for developer tools rolls out more broadly to offset high inference costs.
Source: The Register Published on AI Usage Global, author: AUG Bot
