Rasmus Lerdorf, inventor of the PHP programming language, confirms he is leaving the company. "It has only been a couple of days," he told us by email yesterday. "I really don't know what is next yet... I am enjoying having a bit more time to play with pet projects this week."

Lerdorf, whose widely-used programming and templating system has been especially popular among Web startups, declined to elaborate on why he left Yahoo. But he was blunt about the matter on Twitter this past summer, just moments after Yahoo announced a pact in which Microsoft would power its search results — previously handled by in-house code — while Yahoo would continue to sell ads against the results:

If we had to guess where Lerdorf might end up, we'd lay our money on Facebook, a PHP shop and a fast-growing one at that. The massive social network has no doubt pushed Lerdorf's language to the edges of its performance envelope. More importantly, the young company shows no inclination to outsource software development to one of its largest competitors and turn itself into a second-tier advertising network.

(Pic by Aaron Hockley)

For starters, more Web applications mean more pages running Google-brokered ads. App Engine also runs on Google's preferred programming language, Python, not PHP or Ruby — meaning the developers of tomorrow now know definitively what scripting environment to work in, and the company's talent pool will grow. But ZDnet might have the winning theory: it will make the cost of acquiring startups much lower for Google.

Take YouTube, for example. The company succeeded partly on the merits of being able to keep its databases running. While other video sharing services wilted under their popularity, while YouTube remained online. But once Google purchased YouTube, it had to invest engineers and time into translating the company's backend into something that would work in their other systems. Next time, the transition will be nearly seamless.

So for the startup ecosystem, developers can either go Google or go elsewhere. And if they go Google and build something successful, there will be a threshhold of traffic where they'll be presented with two choices — either pay to play on Google's servers, or sell to Google. In any but the first scenario, Google is all up in your balance sheet. Doesn't look good for Amazon.com and Oracle, which powers Amazon's SimpleDB.