SoftBank is creating a robotics company that builds data centers , and already eyeing a $100B IPO

The $100B IPO target is the tell here. SoftBank isn't positioning this as an internal cost-reduction tool but as a standalone investable asset, which means Masayoshi Son is betting external capital markets will price construction automation as a category, not just a subsidiary function.

This is largely disconnected from recent activity in our archive, as we have no prior coverage to anchor against. It belongs to a cluster of stories about AI infrastructure constraints shifting downstream from silicon to physical buildout, a thesis that has been circulating among hyperscaler analysts but has not yet produced a major standalone company at this scale. SoftBank is essentially trying to create that company before the market does, which is consistent with Son's historical pattern of funding the infrastructure layer beneath a technology wave rather than the application layer on top of it.

Watch whether any of the major hyperscalers (Microsoft, Amazon, Google) sign a construction contract with this new entity within 18 months. A signed customer would validate the market thesis; continued absence of external customers by mid-2027 would suggest this is primarily a SoftBank internal cost play dressed up for public markets.