When discussing growth rates, the time period is critical. Otherwise the entire article could be wrong.
According to a NetworkWorld article on 28 June, Andy Rubin claimed Android had 500K activations per day and grew 4.4% on a week-over-week basis.
And Larry Page claims today in their earnings report that they are activating 550K devices per day.
Taken together, that means Android growth actually slowed between 28 June and 14 July, a period of about 3 weeks. Because x^3 * 500 = 550, therefore x = 1.032, or 3.2% growth week-on-week.
You recognize that the 500K and 550K figures are likely rounded? For instance, when they tweeted 500K, it could have actually been 496K, and 550k could be an actual 553K. Just an example of course. And the latest quoted activation figure could be from a day or two ago. You'd need a bit more information than what's been given to make any solid claim that growth has slowed in the past 2-3 weeks.