The Promise and the Bill: How Cloud Economics Evolved
A decade ago, the tech industry embraced cloud computing with unreserved enthusiasm. The narrative was seductive: shutter your depreciating server racks, fire your data center technicians, and pay only for the exact seconds of compute you consume on AWS, Azure, or Google Cloud.
For early-stage startups with zero revenue and unpredictable usage spikes, this model was revolutionary. However, as digital enterprises matured into steady-state multi-million dollar platforms, CFOs noticed a harrowing trend: cloud infrastructure was consuming 15% to 30% of their entire revenue, expanding faster than gross profits.
The Mathematics of Cloud Repatriation: The 37signals Precedent
In 2022, David Heinemeier Hansson (DHH), CTO of 37signals (Basecamp and HEY), announced that the company was exiting AWS and Google Cloud to install their own Dell servers across two colocation facilities. They demonstrated that their cloud bill was running at $3.2M annually.
By purchasing modern high-density server hardware outright for approximately $800,000 and paying $15,000 monthly for rack space, power, and high-speed bandwidth, 37signals projected to save over $7M over a 5-year lifecycle—while gaining superior compute performance.