Google, Amazon, Meta, and Microsoft could add up to 34 gigawatts of AI compute capacity by 2027 [1].
This massive infrastructure expansion signals a shift into the "age of inference," where the primary focus moves from training AI models to deploying them at scale. The scale of investment suggests that the largest hyperscalers are betting on sustained, exponential growth in AI workloads to justify the costs.
To support this growth, the four companies collectively plan to spend $725 billion in capital expenditures in 2026 [2]. This figure represents a 77% increase from the $410 billion spent in 2025 [2]. The spending is directed toward the global expansion of data-center footprints, and the procurement of hardware [1], [3].
This surge in capacity drives significant demand for Nvidia GPUs and various custom-designed chips [1], [3]. The hyperscalers are scaling their hardware footprints to ensure they can handle the compute-heavy requirements of modern AI applications.
Industry analysts said that the drive for more compute is essential for maintaining a competitive edge in the AI race. By securing vast amounts of power and processing capability, these firms aim to dominate the infrastructure layer of the artificial intelligence economy [1], [3].
“Google, Amazon, Meta, and Microsoft could add up to 34 gigawatts of AI compute capacity by 2027”
The projected spending and power capacity indicate that AI development has moved beyond experimental phases and into a massive industrialization stage. By committing hundreds of billions of dollars to physical infrastructure, these four companies are creating a high barrier to entry for any potential competitors who lack the capital or energy access to compete at this scale.
