AI Capex as a Percentage of US GDP in 2026: What Big Tech Earnings Reveal

Microsoft (MSFT) is projecting roughly $190 billion in capital expenditures for calendar 2026. Meta Platforms (META) has raised its 2026 capital expenditure guidance to $125-145 billion, up from a prior range of $120-135 billion. Together, these two companies are on track to deploy between $315 billion and $335 billion in infrastructure investment within a single calendar year. Against US nominal GDP of approximately $30 trillion, that combined figure approaches 1% of total economic output, a concentration of private capital formation in a single technology theme that earnings calls are only beginning to quantify in full.

Microsoft: Gigawatt-Scale Buildout With Demand Still Outpacing Supply

Microsoft's Q3 2026 earnings call provided the most detailed breakdown of the infrastructure commitment. The company added one gigawatt of capacity in Q3 2026 and is on track to double its overall infrastructure footprint within two years. Management noted the company remains capacity-constrained through at least 2026, indicating that demand for compute is exceeding what the current buildout can serve.

The $190 billion 2026 capex figure includes about $25 billion attributable to higher component pricing, a cost inflation layer that inflates the headline number independent of physical capacity additions. For Q4 2026 alone, Microsoft guided to over $40 billion in capital expenditures for a single quarter, a figure that would annualize to more than $160 billion on its own.

The revenue side provides context for the spending scale. Microsoft's AI business reached a $37 billion annual run rate in Q3 2026, growing 123% year-over-year. That growth rate, sustained on an already substantial base, represents the demand signal management cited when justifying the continued infrastructure expansion. Full detail from the quarter is covered in the Microsoft Q3 2026 earnings episode.

Meta: Raised Guidance and $107 Billion in Forward Contractual Commitments

Meta raised its 2026 capital expenditure guidance to $125-145 billion, up from its previous range of $120-135 billion. The revision, disclosed on its Q1 2026 earnings call, reflected higher infrastructure requirements than the company had initially modeled for the year.

The headline capex figure understates the total financial commitment. Meta added $107 billion in new contractual commitments in Q1 2026 for infrastructure and cloud deals, forward obligations extending beyond the 2026 guidance window that will flow through future periods. The combination of raised annual guidance and $107 billion in new contractual commitments signals that Meta's infrastructure investment is a multi-year structural position, not a single-year allocation.

The Q1 2026 earnings call provides the source data for both the guidance revision and the contractual commitment figure, with additional context available in the Meta Q1 2026 earnings episode.

AWS and Amazon Chips: Revenue Validation at Earnings Scale

Amazon Web Services (AWS) provides the clearest earnings-derived evidence that this capex cycle is generating returns at scale. AWS revenue reached $37.6 billion in Q1 2026, growing 28% year-over-year, its fastest growth rate in 15 quarters, on an annualized base of roughly $150 billion.

Within that total, the AWS AI revenue run rate exceeded $15 billion. Amazon chief executive Andy Jassy offered a comparison to frame the pace: three years after AWS launched, its total revenue run rate was $58 million. The current AI revenue run rate at the same three-year mark is approximately 260 times larger than that early AWS figure.

Amazon's custom chips business adds a further data point on returns. That segment reported nearly 40% quarter-over-quarter growth and carries an annual revenue run rate over $20 billion. Jassy noted that if the chips were sold externally at prices comparable to leading chip companies, the revenue run rate would approach $50 billion. The Amazon Q1 2026 earnings episode covers the full quarter.

Component Costs: A Shared Structural Pressure Across Hyperscalers

One thread running through multiple 2026 earnings calls is component cost inflation, particularly for memory. Jassy described memory costs as having skyrocketed due to insufficient industry capacity relative to demand, a dynamic he said is pushing enterprise customers toward cloud migration rather than on-premises build-outs. The supply shortage, in his framing, is accelerating demand for AWS rather than constraining it.

Microsoft's capex structure reflects the same pressure from the spend side. The roughly $25 billion component pricing uplift embedded in the $190 billion 2026 capex projection means a meaningful portion of the investment total reflects pricing rather than unit volume. That distinction is relevant when interpreting the GDP-percentage framing: some of the nominal spending increase reflects input cost inflation, not a proportional expansion in physical infrastructure capacity.

What the Aggregate Trajectory Indicates

The picture across Q1 and Q3 2026 earnings calls is one of sustained, multi-year infrastructure commitment at scale. Microsoft's stated capacity constraint through at least 2026, Meta's $107 billion in forward contractual obligations, and AWS reporting its fastest quarterly growth rate in 15 quarters collectively indicate that the spending cycle has not peaked.

At the macro level, the combined 2026 capex projections from Microsoft and Meta alone, on the order of $315-335 billion, represent a concentration of private capital formation that approaches 1% of US nominal GDP from just two companies. Factoring in Amazon's infrastructure spending and Alphabet, the total hyperscaler commitment to AI infrastructure represents a measurable and growing share of economic output.

Earnings calls are now providing granular data to quantify what was previously described in broad terms. AI infrastructure build-out is operating at GDP-relevant scale, with multi-year contractual commitments extending well beyond current guidance windows, and revenue run rates on both cloud and chips providing an early read on the return profile.

• Microsoft: roughly $190 billion in 2026 capex, including about $25 billion in component pricing uplift; over $40 billion guided for Q4 2026 alone
• Microsoft AI annual run rate: $37 billion, growing 123% year-over-year; company remains capacity-constrained through at least 2026
• Meta: 2026 capex guidance raised to $125-145 billion, up from $120-135 billion; $107 billion in new contractual commitments added in Q1 2026
• AWS Q1 2026 revenue: $37.6 billion, 28% year-over-year growth, fastest rate in 15 quarters; AI revenue run rate over $15 billion
• Amazon custom chips: over $20 billion annual revenue run rate, nearly 40% quarter-over-quarter growth