Practical: AI & Business News How $93B Turns Google Into an AI Compute Giant
Nov 24, 2025
Google's ambitious plan to double its AI compute every six months is not just a technical challenge but a financial one too, with a massive $93 billion CapEx. The competition in AI infrastructure is heating up, requiring significant investment to stay ahead. Meanwhile, tradeoffs between spending and model efficiency are crucial to watch. Google's new TPU v7 Ironwood promises efficient solutions, while the demand for cloud services reveals a bottleneck in compute capabilities. Sundar Pichai warns of intense competition and the risks of underinvestment in this critical area.
AI Snips
Chapters
Books
Transcript
Episode notes
Compute Must Double Every Six Months
- Google must double AI compute every six months to meet surging demand.
- That pace implies a 1000x increase in 4–5 years, forcing massive infrastructure expansion.
Infrastructure Is The Core Competitive Battleground
- Google sees AI infrastructure competition as the most critical and costly part of the race.
- They aim to deliver systems that are more reliable, performant, and scalable than rivals.
Record CapEx Signals An AI Arms Race
- Alphabet raised CapEx guidance to about $93 billion and plans more in 2026.
- Microsoft, Amazon, Meta and Google now expect to spend over $380 billion collectively this year.
