It depends a lot on the leadership and what are the driving forces behind something like this. Take the Apollo program; you had a charismatic leader whose death had to be avenged by fulfilling his dream. Further, there was a clear vision and a sense of urgency along with new leadership conducive to carrying out that goal. All of that and more made it a success, but the minute that fervor wore off it was killed. Think about it.
Kapil Sibal is one of the most intense, charismatic and brilliant people I know. Yes, he is a politician and by virtue he is suspect, but it still holds true. The programs he set up do work. I am saying this since I benefited from one of them. He started, or pushed off IRIS (http://www.irissciencefair.com/about_iris.php) the indian version of ISEF (http://www.societyforscience.org/isef/). That program is simply amazing, and it gave me the exposure that I desperately needed.
So, maybe you never know. It might work out after all. Whether or not this will change the basic situation of the indian education system is another matter altogether.
Keeping costs lower by be a great technological achievement but will that device be helpful is a question to be asked.
So your estimates line up with that experience. I'd guess that if the Indian government really wanted to provide a lot of these to school students, they could bring the cost down a fair bit by either bulk-buying or bulk-producing, though.
I'd be interested to find out what the intended OS is, and what the plans for providing suitable useful educational software are, though.
(^) For starters, http://www.dealextreme.com/search.dx/search.WM8505
Remember: this is India, you can feed a family for less than two dollars.
But items which aren't produced by local labor aren't necessarily so cheap. For example, the RAM, CPU, batteries, and LCD screen all come from the same places that US manufacturers buy their components. So they aren't going to be any cheaper because the product is sold in India.
The $30 number is doubtful because quite regularly, the government subsidizes projects to meet numerical targets like this. But the article says the $30 cost doesn't consider the cost of distribution. Even if the hardware really costs only $30 to make, distribution is going to at least double that. If this is to be sold by retail vendors, which is the only real way to sell this in a country like India, add another $30 to the cost.