Former Facebook employees launch Coolan, a startup providing predictive analytics for data centers; currently in private beta
Context & Ripple Effects
Coolan is a small bet with an outsized pedigree: founded by ex-Facebook employees, it sells predictive analytics aimed at keeping data centers running efficiently, starting from private beta. That origin matters because Facebook's infrastructure team was then defining hyperscale practice itself — its Open Compute server designs and Big Basin GPU line showed how far operational tuning could push fleet economics.
The arc closed quickly. Within about seventeen months, Salesforce bought the company outright (acquiring Coolan), turning the beta product into an asset of an enterprise software giant rather than a standalone vendor.
First-order effects
- Data center operators in the beta gain access to hyperscale-style failure prediction without hiring Facebook-caliber reliability teams in-house.
- The founders convert their insider knowledge of Facebook-scale operations into a sellable product, exiting via acquisition to Salesforce while still early-stage.
Second-order effects
- Salesforce's purchase signals to other enterprise software vendors that infrastructure analytics is worth buying rather than building, raising the price floor for similar small teams.
- Hyperscalers face a slow leak of operational know-how as alumni productize internal practices, forcing them to weigh openness (Open Compute-style sharing) against losing differentiation.
Third-order effects
- If the pattern holds, data center operations consolidate around analytics layers owned by large software companies, with independent infrastructure-tooling startups functioning primarily as acquisition pipelines.
- Efficiency becomes the contested surface of the data center market — the same pressure later driving Microsoft and startups toward immersion cooling experiments as compute demand outstrips conventional thermal headroom.
The trend: Hyperscaler infrastructure know-how is spinning out into analytics startups that get absorbed by enterprise software buyers, making data-center efficiency a software-owned layer.