Projected AI CapEx to Reach Approximately 2 Percent of US GDP by 2025, Still Significantly Lower than 19th Century Railroad Investment

This is not investment advice. The author does not hold positions in any of the stocks referenced in this article.

AI Expenditure: An Economic Powerhouse

The surge in Artificial Intelligence (AI) investment is undeniable, with expenditure ranging from substantial sums allotted by major tech firms—often referred to as hyperscalers—to smaller financial commitments by mid-sized businesses. Individuals are also contributing to this trend through various AI subscription services provided by companies like OpenAI.

The AI infrastructure build-out is so gigantic that in the past 6 months, it contributed more to the growth of the U. S.economy than /all of consumer spending/The ‘magnificent 7’ spent more than $100 billion on data centers and the like in the past three months *alone*1/🧵 pic.twitter.com/sHMK1zI0sP

— Christopher Mims 🤌 (@mims) August 1, 2025

This significant trajectory of AI expenditure is fundamentally reshaping the U. S.economy. For context, the “Magnificent Seven”hyperscalers, including Meta, Microsoft, Google, Amazon, Apple, Tesla, and NVIDIA, collectively invested over $100 billion in AI-related infrastructure in just the last quarter.

Hyper-Sclarers’ Investment Trends

There appears to be no slowdown in sight. Notably, Meta has projected its AI capital expenditures to reach $69 billion in 2025, with indications that it could surpass $100 billion by 2026.

Meta’s reply to Stargate comes through Prometheus at 1 GW and Hyperion at 2 GW, running multi-billion-dollar GPU clusters that sit in “tents”.- Meta – Prometheus IT Power by end of 2026 is 1, 020MW. The total number of chips used is 500, 000. Total compute power is 3, 171, 044, 226… pic.twitter.com/FLtiAMDk3F

— Rohan Paul (@rohanpaul_ai) July 14, 2025

Among the innovations, Meta plans to establish a gigawatt AI cluster known as Prometheus, equipped with up to 500, 000 NVIDIA GB200/GB300 AI GPUs. Additionally, a more ambitious 2GW AI cluster named Hyperion is on the horizon.

Investment Plans of Other Hyperscalers

Microsoft forecasts an expenditure of $30 billion in AI capital for the upcoming quarter concluding in September, with an astounding $100 billion budgeted for the new fiscal year. Amazon, too, has set its sights on $100 billion for 2025, while Google/Alphabet anticipates an investment outlook of $85 billion for the current year.

Moreover, a colossal $500 billion AI mega-project called Stargate has enlisted numerous firms, including OpenAI, SoftBank, and Oracle. Recently, OpenAI secured a remarkable 4.5GW data center agreement with Oracle, which entails annual payments of $30 billion commencing in three years. Sam Altman, OpenAI’s CEO, has indicated that establishing a 5GW data center capacity typically requires around $100 billion in investment.

The expansion strategy is further bolstered by OpenAI’s arrangement with Oracle in July to develop an additional 4.5GW of data center capacity, although the exact location for this expansion has yet to be disclosed.

OpenAI has also made a deal with CoreWeave, augmenting its existing contract with an additional $4 billion on top of $11.9 billion for access to CoreWeave’s extensive GPU resources. Impressively, OpenAI has signed data center agreements totaling $100 billion this year alone, aligning with Stargate’s investment goals for 2025.

Economic Implications of AI Capital Expenditures

Currently, approximately 250 data centers are under construction across the United States, pointing to a burgeoning AI CapEx landscape that is making a significant contribution to U. S.GDP growth.

Veteran consultant Paul Kedrosky provides a theoretical framework to visualize the macroeconomic implications of this exceptional spending activity (details here).Starting with NVIDIA’s annualized data center revenue projection of $156 billion for fiscal year 2026, he estimates NVIDIA’s share of total AI CapEx in the U. S.to be around 50%.This suggests an overall CapEx for 2025 of $312 billion.

Applied with a multiplier of 2, this translates to a positive impact on U. S.GDP of approximately $624 billion, indicating that AI CapEx will constitute around 2.08 percent of the U. S.GDP.

Historical context is essential here; peak railroad investments accounted for approximately 20 percent of U. S.GDP in the 19th century. This suggests that the current AI CapEx boom has substantial potential for further growth before it approaches the historical spending levels seen during the railroad expansion era.

Source & Images