A company that doesn't make money buys another company that doesn't make money. The Silicon Valley economy exists entirely on its own with no relation to the "real world".
In the past 12 months Twitter has built out an entire north american sales force lead by former FIM exec Adam Bain, who is v good at what he does. So those numbers are before any international expansion, and do not include API licensing (via Gnip)
It is very disingenuous and flat out wrong to write Twitter off as having no revenue. They have no profit, but the point of a company at their stage is not to book profits, it is to spend as much money to grow as quickly as possible (building that sales team alone isn't cheap - let alone infrastructure). If they were making money, the investors would sure as hell be asking why. This is simple startup economics where you have a company and product in hypergrowth.
Twitter is so ingrained in the mainstream conscious now. My non-tech family members and friends kept mentioning it all weekend in context of the royal wedding and the bin laden death. The cable news networks spent a lot of time just showing their viewers what is on twitter
It is here to stay at least for a little while, so time to throw out the 2009-era skepticism and false criticisms
As an aside, Tweetdeck are also healthy revenue wise. Healthy enough to be an 18 person company that last raised a small round a year ago and that can charge $50k per ad partner for placement[2]
[1] http://www.businessweek.com/news/2011-01-24/twitter-s-ad-rev...
[2] http://techcrunch.com/2009/06/01/want-your-service-integrate...
The thinking was that Gnip would implement the client once, and then expose a standard interface so that a dev could integrate with all of these services using just a single API
The second part was that instead of polling, Gnip would push update data to you using XMPP or callbacks.
I think they found that there isn't that large a market for apps, at that time or this time anyway, that needed to integrate with a dozen or more services to make paying Gnip worthwhile
The second part solved itself through either pubsubhubbub or API's implementing their own callback.
Gnip downsized for a while, but saw a resurgence with the Twitter deal. It makes sense for Twitter to offload serving the API, implementing track, etc.
I think datasift is a bit different. Actually a bit more advanced since they support many more ways to slice and dice the data, analytics, and more. That space is more like hootsuite and the similar services - you want to mine social network data and extract key analytics. Gnip is more about making API consumption for apps easier.
Gnip has 50+ services, Datasift about 6 or 7 - so you can see where they are beginning to diverge.
When Gnip first launched, I spent a day with the team as part of covering the launch for techcrunch. My own conclusion at the time was that Gnip was a stop-gap measure between then and whenever the protocols such as PuSH would be stable enough to see wide adoption. There was also the element of outsourcing your API which didn't seem right - Mashery does something similar with outsourcing dev community, API tokens etc. But Gnip have proven otherwise with the twitter deal. (edit: for reasons why twitter would use them, see [4]
What could be interesting is Gnip has morphed into part of the old Gnip, with aggregation and normalization, plus something akin to Cloudflare but for API's - while datasift has become a datamining and analytics platform more for internal business use.
The space that Datasift is in will be huge. With Gnip I am still a bit indifferent only because I am implementing an app at the moment that talks to a dozen or so different API's and I couldn't see how I would use it or why I would pay for it, but the twitter firehose isn't one of hte API's (but twitter auth is - and Gnip, from what I know, doesn't touch auth at all)
As an aside: and interesting point in history is that when Gnip launched twitter totally rejected them, only later to give them access and then take it a step further with a full commercial partnership[2][3]. that was back when the twitter firehose was 5-6 tweets a second, it hit 5k a second yesterday
[1] http://www.techcrunchit.com/2008/06/26/the-new-datastream-ag...
[2] http://www.techcrunchit.com/2008/07/18/battle-over-twitter-o...
[3] http://techcrunch.com/2008/07/18/twitter-plays-nice-xmpp-fir...
[4] http://www.readwriteweb.com/hack/2010/11/why-is-twitter-part...
1) No ads. I don't see any now. "But, they can add advertisements later!" No they can't. Adding advertisements to Twitter will destroy their brand and everything that makes them "cool."
2) No clear picture on their profits and revenue. Profitable companies or companies with a solid revenue model have nothing to hide. There's no reason to hide your money situation if you're successful or have a successful strategy. Twitter doesn't. Their strategy relies on selling to the VC community, not selling to users.
3) Buying up competitors before you're posting profits means twitter's only asset (users) are hard to maintain control of.
Promoted Tweets.
2) Promoted tweets, api firehose. Their tries at adding ads to twitter certainly sucked, though.
Perhaps on Tweetdeck, as a random example.