This was the first startup I worked in. And it was also where I received my first ever paycheck from software work, which felt significant at the time.
The idea was straightforward: university computers sit idle at night. Thousands of them, in labs and offices, powered on and doing nothing after faculty leave. What if you could use that collective spare capacity as a distributed computing platform?
The plan
We built a BOINC-based system — BOINC is the open-source framework that powers things like SETI@home and Folding@home. The setup was: install our software on the university's PCs, and after hours, when no one was using them, the machines would start processing whatever task had been submitted to the platform. The idea was to sell that computing capacity to anyone who needed it — scientific computation, rendering, anything batch-workload-shaped.
Why it didn't work
The machines were old, which meant the aggregate compute was less impressive than it looked on paper. Maintenance overhead was real — managing software on hundreds of heterogeneous machines with no guaranteed uptime is not a small problem. And fundamentally, the economics didn't work: the cost of maintaining the network was higher than what you could realistically charge for the compute it produced.
So the startup failed. Officially. I don't think of this one with regret — it was early, it was naive in exactly the way first startups tend to be, and it taught me that the gap between "technically possible" and "economically viable" is worth understanding before you build.