CloudShare

The company's own team page contains a sentence most companies would not print. Guterman, it says, purchased CloudShare from its investors for four dollars, and then built it into the market leader in laboratory environments. The company had raised around $26M across three rounds. A business does not invent a story in which its own investors valued it at four dollars, which is the best reason to believe it - and the ending is that it was sold to a Denver asset manager at a reported $60M to $80M after roughly a decade of bootstrapped, profitable growth.

That sequence is worth sitting with, because it inverts the standard narrative this timeline records over and over. Raise money, grow fast, sell. Here: raise money, have the investors write it off entirely, buy it back for a nominal sum, and then grow slowly on your own revenue until it is worth tens of millions. The venture route is the one that gets written about, and it is not the only one that works.

The technical problem is specific and is one this site's readers will recognise instantly. Enterprise software assumes a data centre: several machines, particular network topology, licences, an installed state that takes a day to reach. Anyone who has taught such a product knows the real constraint is not the material - it is that thirty students each need a working copy of an environment that takes hours to build and that they will inevitably break. CloudShare's answer is to snapshot a fully configured environment and hand every user an independent replica in about a minute.

The insight in the business model is that a training lab and a sales demonstration are the same artifact. Both are a working replica of a system the audience does not yet have, in a state that shows it doing something useful. The difference is what the audience is asked to do afterwards, and that difference is entirely in the framing rather than the infrastructure - which is why one platform serves training, presales demos and proofs of concept.

Which is where this entry meets an argument already on this site. A proof of concept is supposed to be a test that can fail; a demonstration is supposed to succeed. When both run on identical infrastructure, prepared by the same people, the line between them is a matter of intent rather than of setup - and a proof of concept designed so it cannot fail is a demonstration wearing a different name. Good lab platforms make honest testing easier and dishonest testing easier by exactly the same amount.

Customers include Palo Alto Networks, Atlassian, HP, Microsoft, RSA, Salesforce and ForgeRock - several of which appear elsewhere on this timeline, which is the ordinary condition of the layer this cluster documents: the companies you have heard of run on companies you have not.

And it closes the cluster where it began. Kryterion partners with CloudShare to add proctoring to hands-on labs, which is the industry converging on the hybrid: the realism of a practical assessment, in a reproducible environment, with the identity verification of a supervised exam. Every piece of that sentence is a different company, and none of them writes the questions.