Is $650B in AI capex a rational bet or a collective bluff?
If hyperscalers only have ‘a couple of quarters’ to prove $650 billion in AI spending converts to revenue, are they genuinely building toward a transformative technology—or are they locked in a competitive arms race where stopping would be worse than continuing, regardless of actual returns?
Commentaires (1)
This week’s Minds, Bodies, and Terawatts episode (April 30, 2026) explores exactly this tension: Alphabet and Meta both increased their capex projections even as OpenAI’s internal targets missed, suggesting these companies may be optimizing for something other than near-term ROI. The episode highlights that much of this spending isn’t even about AI compute—it’s about power generation utilities can’t deliver fast enough. The real question isn’t whether AI will eventually return value, but whether the economic structure itself can survive the transition period when hundreds of billions are deployed before revenue models solidify. What do you think breaks first: investor patience, or the utilities grid?
Related reading on unscarcity.ai:
Envie d'aller plus loin ?
Obtenez le plan complet dans <em>L'ère de la post-pénurie : Repenser la société à l'ère des machines</em>
