# Spending too much on AI? **URL:** \ **Category:** Macro Economic Trends and Risks\ **Created:** [November 26, 2025, 3:49pm UTC](https://discussion.fool.com/t/spending-too-much-on-ai/122309 "2025-11-26T15:49:48Z")\ **Posts on this page:** 3\ **Page:** 2
**Author:** ![Leap1](https://sea2.discourse-cdn.com/motleyfoolfree/user_avatar/discussion.fool.com/leap1/32/72176_2.png) [@Leap1](https://discussion.fool.com/u/Leap1)\ **Post date:** [December 6, 2025, 12:39pm UTC](https://discussion.fool.com/t/spending-too-much-on-ai/122309/21 "2025-12-06T12:39:16Z")
> [@JLC](#): > > AI build out is only 0.8% of GDP. AI Overview AI-related investment is estimated to be **around 1.3% of U.S. GDP in 2025** , with some estimates for total AI buildout spending reaching as much as 2% of U.S. GDP. This spending has been a major driver of recent U.S. economic growth, accounting for a significant portion of GDP expansion in the first half of 2025. Key Details on the AI Buildout and GDP - **Investment Share of GDP** : Current estimates place the share of U.S. GDP dedicated to AI-related capital expenditures (data centers, servers, chips, and software) at approximately **1.3%**. Some sources suggest it could reach up to 2% for 2025 alone. - **Contribution to Growth** : The _impact_ on GDP growth is even more substantial. In the first half of 2025, AI-related investments were reported to have contributed as much as 0.5 to 1.1 percentage points to U.S. GDP growth. One analysis even suggested that 92% of U.S. GDP growth in H1 2025 came from AI-related data centers and supporting technology investments. - **Comparison to Past Booms** : While the current AI investment boom is massive, its share of GDP is still considered smaller than the peak of the late 19th-century railroad investment boom (around 6% of U.S. GDP) or the peak of the dot-com era’s telecom buildout (around 1.2% of U.S. GDP, but the AI boom is currently on track to exceed this pace). - **Global vs. US Focus** : Most available data focuses heavily on the U.S. economy due to the concentration of major tech companies (the “Magnificent 7”) driving the buildout. Global spending data is less precise as a single percentage of world GDP, but it is a major factor in worldwide economic projections. In short, while the total percentage of GDP is relatively small compared to entire economic sectors, the _rate of growth_ and its _contribution to overall economic expansion_ are remarkably high, making the AI buildout a critical economic driver in 2025. ---
**Author:** ![Leap1](https://sea2.discourse-cdn.com/motleyfoolfree/user_avatar/discussion.fool.com/leap1/32/72176_2.png) [@Leap1](https://discussion.fool.com/u/Leap1)\ **Post date:** [December 6, 2025, 12:44pm UTC](https://discussion.fool.com/t/spending-too-much-on-ai/122309/22 "2025-12-06T12:44:29Z")
> [@MarkR](#): > > For large investments, it’s usually a good idea to estimate what the ROI is going to be. I’m pretty sure all companies do this when determining where to place their capital. Sometimes they are correct and sometimes they are wrong. Speaking to a young manager three weeks ago. I asked him what regional is telling him about the economy. He is as close as I can get to regional. He said they are reporting nothing to him. My response off the cuff, “There is nothing factual to report. We are talking economic theory. But the recent cutting is because they need to gear for the first time in years to worsening conditions”. Mark it is the regearing that is the problem. Company managers trying to hold onto ROI will gear down their expenses. This is a house of cards. ---
**Author:** ![Goofyhoofy](https://sea2.discourse-cdn.com/motleyfoolfree/user_avatar/discussion.fool.com/goofyhoofy/32/4721_2.png) [@Goofyhoofy](https://discussion.fool.com/u/Goofyhoofy)\ **Post date:** [December 6, 2025, 2:38pm UTC](https://discussion.fool.com/t/spending-too-much-on-ai/122309/23 "2025-12-06T14:38:05Z")
> [@MarkR](#): > > For large investments, it’s usually a good idea to estimate what the ROI is going to be. I’m pretty sure all companies do this when determining where to place their capital. Sometimes they are correct and sometimes they are wrong. At Westinghouse we had a program called Vabastram (Value Based Strategtic Management) which “told you” whether an investment was worthwhile or not. You input the cost, an assortment of variables (interest rates, hurdle rate, capital requirements, etc.) and an answer would pop out, almost like in the cartoon computers of yesteryear. It took us about 12 minutes to realize that by manipulating the inputs you could make it produce any answer you wanted, which was usually “yes.” Occasionally you’d have to stretch reality so far as to convince you the answer really was “no”, but even then if you really wanted a project you could make it work. Westinghouse, at least in that incarnation, is defunct. I’m sure somewhere Vabastram - or a derivative or parallel - still exists, and highly paid executives are using it to justify the decisions they’ve already made. [Previous page](https://discussion.fool.com/t/spending-too-much-on-ai/122309.md?page=1)