Data center investment tied to the spread of generative AI has reached a cumulative scale exceeding $3 trillion, with Big Tech raising funds through large bond issuances. The simultaneity of the spending and the uncertainty over monetization have fueled debate about its sustainability.
AI Infrastructure · The Capital Wave
The $3 Trillion Race to Power AI
AI data-center investment is projected to top $3 trillion by 2030 — an "enormous, simultaneous, foundational, speculative" buildout. But money may no longer be the bottleneck: power is.
$3T+
Data-center investment through 2030 (Moody's)
$5T
Total scale incl. $1.5T IG bonds (JPMorgan)
~$750B
2026 capex, top-14 firms (BNEF)
$25B
Meta AI bond issuance, April 2026
Annual AI capex more than doubles
Goldman Sachs projection — yearly AI-related spending, 2026 → 2031
Roughly a 2× rise in annual spend over five years.
Power is becoming the real constraint
As capital floods in, electricity — not financing — is emerging as the binding limit. AI performance is starting to look less like a software problem and more like a question of thermodynamics.
4% → 9–17%
US data-center share of electricity, 2023 → 2030
+905 TWh
Additional demand by 2030 (Goldman Sachs)
$11M / MW
Cost to build 1MW today — double the 2022 figure
The Buildout Case
Concrete demand: power-layer firms like GE Vernova and Vistra report surging AI-driven backlogs
Global race — China readies a ~$295B (¥2T) nationwide buildout
Funding diversifying from cash flow into bonds and private credit
The Bubble Question
"Boom, Bubble, or Buildout?" — US private AI investment hit $285.9B in 2025
Speculative power plans (SMRs, fusion) and construction-labor shortages
Risk of heavy dependence on a single customer
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…