They're basically a consulting company with in-house platforms for data management. You pay $100k+ per core pricing for their analytics platform, then you pay thousands more for implementation 'ninjas', consultants, training etc. It's a mix between a services and a technology company targeted at government accounts.
It's funny, because every time someone posts a news article about Palantir, this is the top comment. And every time, the replies to that comment do a pretty good job describing what it is they do:
A patent for a long left-click to show a context menu when the right mouse button for some reason isn't available? Sheesh. Some of these are enough to turn anyone into an anti-software-patent crusader.
Maybe I'm not remembering it right, but when macs had one button only and no fancy trackpads, you could hold it to do a right click (you could also just ctrl+click)
The main use case I’ve come across is network analysis. After awhile in Iraq, the military realized killing people planting roadside bombs was stupid and ineffective since these were low level individuals sometimes doing it just for money to feed their families. The main focus shifted to building up a picture of larger enemy networks so they could be taken down in a more holistic fashion. McCrystal built the fusion cell concept [1, 2], bringing in intelligence from multiple agencies to build target packs of people and networks for operators to take down. I’ve never used the software, but I believe a key use was to help various agencies surface connections in the troves of data to build these target packs. Later, someone from the McCrystal consulting group [3] said they had a former CIA analyst who used the software to find hidden connections between people in a company, to uncover and reward the critical individuals who get things done outside the normal organization hierarchy. So again, network analysis, but directed toward finding super star employees instead of terrorists. I’m sure the software is more general then this and has other multiple other use cases, but this is the best of my second hand knowledge and seems to go along with what others have linked to here describing a node and vertices analysis tool
See https://news.ycombinator.com/item?id=7294743 and Thomas' reply. That's the extent to which Palantir was "involved" with HBGary Federal. (And while it should go without saying, I'll say it anyway: There was no attempt to infiltrate and destroy Wikileaks. Just a lot of hairbrained talk.)
Why should it "go without saying"? Have the facts been laid out and proven in an exceptionally clear and public way? I don't think so.
And "hairbrained" is an strange adjective to use for a company that is almost 100% engineers from top schools, pulling in a billion in funding, and working for the CIA.
All I know is the contents of the leaked presentation that laid out the plans against Wikileaks. I will ignore any PR and damage control about "hairbrained talk".
Right. All due respect to Thomas, but downplaying/minimizing literally every single shady thing the government does is par for the course for him. As his recent concession on Dual_EC_DRBG demonstrates, sometimes the government really does engage in damaging, moustache-twirling villainy.
Correct, In-Q-Tel wrote initial checks for Google and Facebook. Someone figured that instead of having a fancy logo of an octopus reaching out over the globe of earth (or an eye above pyramid), its much better to just found a company with fancy logo with colored fonts, or an F, and make them harvest intel instead.
Only if we got a chance to look at their emails à la Hacking team style… Maybe they'll be the first to join the 1TB+ data dump club to go along with all their rounds of funding ;)
Sells crap to CIA. You have all these targets? OK, input that data in our system then we'll show you have they are connected by a graph, now pay us $10M.
If it's unlikely that they'll go public, how do the investors hope to get their money (and more) back? I know there are other ways, but I don't have a feel for which kinds of arrangements would really be likely.
I don't see why they wouldn't, but the OP says this:
"Palantir’s next step isn’t likely to be filing for an IPO as founder Alex Karp hasn’t been keen on that idea in the past"
After $2.32B in funding it hardly seems likely that the plan is to go another round and then IPO. Why would Karp be any keener on it then than now? That leaves the possibility of no IPO at all . . . hence the question.
That induction does not hold. Karp hasn't been keen on an IPO for the same reasons why many other private companies have been of late -- because it's easier to focus on growing the company without the ever-increasing overhead of being a public company. That in no way implies (or even suggests) that they never want to IPO. I used to work at Palantir, and I've never heard anything to indicate that they want to remain private forever. Even early on, Karp talked about not wanting to IPO (or sell out) too quickly. His plan was always to aggressively grow Palantir into an incredibly successful (and valuable) company, and only then to exit.
You give up some control of the company or enough control that shareholders own a majority stake, which is likely the case with the companies you mention. You also are subject to a boatload of new regulations and reporting requirements. Going public is a way to raise lots of money. It doesn’t really give a company any tangible benefits. Especially recently, companies that have gone public have ended up being valued at less than their last funding round.
In the same way that a non-dividend paying stock (such as BRK) is a Ponzi Scheme. Furthermore, there is an active secondary market for shares in Palantir. It's obviously far less liquid than the stock of other similarly sized public companies, but it's still an equity investment.
Palintir is one of those companies that strikes me as one that really never goes public. If someone ever built The Machine, I would expect it to be them.