Applied Digital (APLD)

The sleeper story is that APLD is morphing into a power aggregator and execution specialist while everyone focuses on their data center buildout; they're seeing 50+ stranded power sites weekly and becoming the scarce capability that can actually build for hyperscalers. The $5B hyperscaler win validates the model, but the 33% gap between cash collected ($8M) and revenue recognized ($12M) while DSO metrics vanish entirely suggests collection problems brewing beneath the headline growth. The binding constraint isn't power or demand, it's their ability to execute on multiple sites simultaneously, which Cummins essentially admitted: 'I'm thinking less about the demand side...more on how much can we build.' Most revealing quote: 'Nothing is done until it's done' regarding the 900MW pipeline after three months of 'advanced discussions' on Polaris Forge 2. My view changes if they actually close a lease in January and show positive operating cash flow.