Did One Laptop per Child truly fail as a startup? It was always a nonprofit, so of course it didn't make money. While the technology was lacking in many respects, it did serve a lot of children at the time, and proved the market that Chromebooks came to later dominate with a similar approach (linux-based OS with minimal resources at $100-300 price point).
It's a pretty classic example of a CEO who was very capable at establishing business partnerships but had such a rigid view about the problem they were trying to solve. They never really proved that the 1:1 concept actually had merit, but ended up being so invested in quantity over quality.
In developing countries, the computers were too simple to teach effective computer skills. And in undeveloped countries, the laptops did nothing to address actual education problems.
Chromebooks have the benefit years of Chrome apps being developed to solve actual problems. They also give up a lot of the goofy ideas like hand crank recharging or mesh networking and etc.