FEAR & GREED | Business News Q+A: $US700 billion on AI, but profits may be years away
Feb 23, 2026
Dr Kevin Hebner, Global Investment Strategist with macro and tech-cycle expertise. He breaks down the US$650–700bn AI buildout and where that cash goes. He explores capacity constraints like power, chips and talent. He explains why tech waves often take decades to pay off and why reliability hurdles can delay real returns.
AI Snips
Chapters
Transcript
Episode notes
Scale Of AI Capital Spending
- US tech AI spending this year is enormous, roughly US$650–700 billion and about a third of Australia’s GDP.
- Kevin Hebner says much of it is AI CapEx: data centres, GPUs, racks, cooling and connectivity to train and run models.
Bottlenecks Threaten The Buildout
- Rapid AI buildout faces physical and supply-chain bottlenecks like power, turbines, engineers, chips from TSMC and EUV machines from ASML.
- Hebner warns single-vendor dependencies create constraints on scaling data centres and semiconductors.
Tech Waves Need Decades Not Years
- Major tech waves historically take decades to deliver practical, profitable products, not years.
- Hebner compares current AI (three years after ChatGPT) to 1999 Internet early days, so real returns may be much further off.