Every hyperscaler earnings call now leads with capex figures that dwarf what AI products actually bring in. The gap is widening, not closing. Infrastructure may be a moat, but investors eventually want a payback timeline. The enterprise side is stranger still: pilots everywhere, production deployments rarer, and falling per-token pricing cuts both ways. Cheaper inference should drive volume, yet it compresses margins for anyone selling on top of it.
AI capex is outpacing revenue — how long can this hold?
Replies
No comments yet — be the first to share your thoughts.