Palantir raises $680M

You don't let your key spy software company fold up.

Series Q. Quite appropriate in a James Bond spy movie kind of way.

I don't think that, strictly speaking, Palantir "needs" more cash. However, it puts them in a more secure position grow aggressively, even if markets take a hit in 2016.

Contrasting them to Atlassian, Palantir is much more ambitious. While they are best known for their government work, they are now far more than a government contractor. They have very strongly expanded to into many different other fields, and aim to be the data analysis leader, period. In that sense, they are a more scalable business.

To be clear, I'm certainly not saying that Palantir's approach is any way superior. It is certainly more risky. My point is that, while Atlassian is steadily profitable, Palantir is pursuing a path of "Go big, or go home." Raising more money lowers the chance of "going home."

(Disclosure: I worked at Palantir from '06-'08. Everything I post about them is my own opinion based on publicly disclosed facts.)

How do you know Atlassian has a much more scalable business, without knowing the specific details of Palantir's business operations, its growth rate, its profitability, its revenue, etc?

Besides, what's bigger than spying and gathering data on citizens these days? Governments around the world are allocating increasingly vast sums of money to it. It's certainly a drastically larger business opportunity than Atlassian's $319m in sales (after 13 years). Governments are not going to spend less on what Palantir does, they're going to spend a lot more, and then after that they're going to spend a lot more. What would it be worth to be one of the principle software contractors for the US Government for the next 20 years, and perhaps the five eyes in general? How about a $60 billion market cap like some of the other big defense contractors? Why not... that's the age we're entering.

They also raised $605.49 in 2014. Total funding to date is $1.8B over 17 rounds. At a $20B valuation this puts them at about 11:1 valuation over funding, which is informative about their investors guess into their growth prospects.

https://www.crunchbase.com/organization/palantir-technologie...

I don't know, I'm not seeing a "sleeping giant" here... Plantir has been around for a long time already at this point- Getting an additional cash infusion, especially one so small for a company of Palantir's size, sounds like the move of a company that is struggling.

> Data is the oil of the 21st century.

Interesting. So you're saying we'll reach "peak data" in about 100 years?

I disagree with the analogy. Human population will not grow from 1 billion to 7 billion because of data, that is the kind of change we got with oil. Oil is a hard asset, data is not entirely hard asset.

Everyone in tech knows about Palantir.

The snake oil?

As someone who has worked with Palantir and used their products in various, indirect and proxy contexts it gets confusing as to what you can talk about and what you can't depending on god knows what 15 papers you signed to be on the contract which has access to Palantir.

I suspect this is a big part of why people don't talk about it much and why their website is so incredibly vague and, honestly, really terrible at explaining what they do. I found this Quora question and answer which ultimately doesn't really describe what I have interfaced with before https://www.quora.com/What-specifically-does-Palantir-do. Perhaps I simply didn't interact with that part of the product.

Sorry for the semi vague and terrible answer. Briefly looking at some of these videos may be of help as it shows Palantir and other software doing some sort of entity extraction from documents https://www.youtube.com/watch?v=3NAVSyfjcDo

Palantir has a core team of engineers building out a very general data analysis framework. Their core product consists of a GUI for manipulating and sharing annotated graphs (i.e. edges and vertices). The actual edges and vertices of the graph can have complex types defined by some arbitrarily detailed ontology. The GUI also includes tools for producing visualizations and charts from numerical and textual attributes of the edges/vertices. The core product is very powerful when configured properly and it is used to solve important problems in both government and industry.

Palantir also has a much larger corps of consulting engineers (FDEs in Palantir parlance) who actually build the ontologies, import a customer's data in to Palantir, and then provide ongoing support to users. A huge portion of this work is writing scripts to munge the data from arbitrary formats (SQL tables, CSV, Excel, etc) to the format supported by Palantir. A core reason for Palantir's success is that they convince large numbers of top engineering students from MIT, Stanford, Berkeley, Harvard, etc, to do this relatively unglamorous work for below-market pay.

Due to the imbalance in headcount between core engineering and the consulting division, Palantir, in practice, operates more similarly to Accenture or McKinsey than to Google or Facebook. However, for some unknown reason, their valuation seems to be calculated like that of a high margin software company, rather than like that of a low-margin services business. In particular, to increase revenue by N dollars, Palantir needs to hire O(N) employees.

TL;DR: it's the McKinsey of this century.

From my relatively limited experience in the space: most companies have trouble actually doing anything with their data beyond slicing it a couple of ways [0] and making bad graphs with it; these same companies also have a lot of trouble hiring and retaining [1] people capable of doing it (I know because they end up my clients).

Palantir will offer large amounts of money (despite accusations of below market pay; although a good portion is in equity) and a half-decent lifestyle-wise, not too corporate environment to young smart engineers to go and help them outsource that function for the kind of companies that do not care about paying large fees to "get the best", and for whom the consultant status helps cut through the red tape.

The "products" are, from my limited understanding, more like a loose collection of accumulated knowledge from clients which can help future deployments not reinvent the wheel (particularly since so many of the engineers are young and inexperienced). This is also a selling point. "We get it right because we did it before with all these prestigious clients".

Each client has its own infrastructure and schema, and so most data related projects are naturally oriented towards a lot of manual effort - talking to technical staff, reading through code, understanding the data customers and feedback loops... and whilst many companies in the space pretend to offer "plug and play" BI (or data science, or whatever) in practice without a strong data team behind it, these end up failing pretty badly. So I see Palantir's competitive edge as its staff, not its products, and I don't think much (as a data professional) of pure product data companies.

In many ways, Palantir is what McKinsey was half a century ago: it offers job variety and a pick of the best clients to the "top ranked" fresh grads and early professionals; it offers these "guaranteed fresh" minds to clients who have trouble attracting them in-house to solve difficult problems that can't in theory be solved by the internal talent pool; it brands itself as the "top" company globally doing this. If it is successful, it will just keep growing its competitive advantage ("you don't get fired for hiring IBM"). And I think their focus on people instead of software is a great idea in an age of exceptional open source products and ever cheaper hardware.

Where might its cash flow come from? I see two sources:

a. as you get further into a data project, you've built the infrastructure and it is comparatively easy to provide answers. For example, once I'm done with a data warehouse, I can produce a complex report or query answering just about any question management can throw at me in 10 minutes to 2 hours; this will take them two weeks to use. So the time actually spent working, vs the monthly retainer, just keeps going down with time (and of course experience - if you wrote the query for another client you can just change the variable names and output).

b. the value add from getting the data right is extremely high. Cf [0] below. For many modern companies it can be the difference between a 50% increase in profits and not being a going concern the following year. As CEOs realize this, and realize that no, you don't get "80% of the quality for 20% of the price" by going for the lowest bidders, prices track up and stay there.

[0] Most classically trained managers that I've met are unable to cope with more than 2 variables at the same time. A graph is basically a 1D or 2D model with your brain and past experience being the curve fitting. This becomes an issue if the business model fits 20 or 100 dimensions better (e.g. many trading strategies and even market making) and most of your competitors are very comfortable with it.

As an example, I've witnessed two e-commerce companies in the same country with a 10x difference in conversion rate; one had statisticians to manage its marketing spend, the other preferred going by the traditional way of "if its conversion rate is high I'm going to add more budget".

[1] I know an exceptional data scientist who changed jobs 4 times in the last year... each workplace bidding him up!

They build two different, non-overlapping products (or at least they did when I last looked into them about 3 years ago).

1) A collaborative semantic graph tool [1] -- this is the thing they're best known for. The client tool is a unified interface for both adding new information to the semantic graph as well as searching/exploring what's already in the graph. It does this in a way that scales to a fairly large number of simultaneous users and allows organizations to build and explore collective "knowledge". The input focus is on manually selecting and "tagging" semantic elements in free-text, aligning those with the system's ontology framework and even connecting together different semantic elements with certain constrained relationship types. [2] Over time this aggregates into a hihgly curated knowledge-base that can be mined more efficiently than reading lots and lots of reports and news articles.

They support a few other odds and ends (maps, various statistical tools on the semantic graph, different search methods). There's plenty of other vendors that do parts of this, but Palantir brought excellent user interface design and better collaboration tools to the party.

These days, their core semantic graph tech could probably be replicated by half a dozen engineers using HTML5 technology and some sort of scalable hadoop-like back-end in a few months. It's not that interesting of a technology. On a personal level, you could do...~70% of what they do with something like yED or Visio. Protege [3] would probably get you another 10% of the way there. A good google docs-style collaboration another 10% and so on. There's nothing uniquely revolutionary about the tech other than they brought it together in the right way.

My impression is that most of the money they make is on integrating customer data sources into the tool so users have more things to markup and add to the knowledge base. Enterprise integration is a very lucrative and time-consuming job and a competitor's quick rewrite (like I mention above) won't be able to compete against Palantir's entrenched relationship and large marketing teams. This is where most of their engineering resources are going at the moment.

2) A "big data" statistical exploration tool meant for finance. [4] I don't know much more about it other than they designed a new language for it and if connected up to the right kinds of data sources looked really powerful in the videos I saw of it in action.

They have lots of open positions all over the world, which makes me think they have lots of offices all over the world, which means they do lots of sales and marketing globally, penetrating new markets. My guess is that the huge numbers of enormous funding rounds go into marketing pushes into different verticals and locations.

They used to have a demo on their website you could play with, but they've taken it down. The tool was nice looking for 2010-ish, but it had a Java JRE requirement that was pretty old and wouldn't run on newer runtimes.

1 - https://www.youtube.com/watch?v=f86VKjFSMJE

2 - https://www.youtube.com/watch?v=l-UrRJMDGWc

3 - http://protege.stanford.edu/

4 - https://www.youtube.com/watch?v=U8YMPS_DZPk

I don't know about the databases, but I know they use a lot of Java. I think they might have their own database.

For frontend they use TypeScript: https://www.palantir.com/2013/12/announcing-three-open-sourc...

They don't seem to think that is a good idea: http://www.cnbc.com/2014/03/19/free-advice-dont-go-public-sa...

Yes. Their software analyses connections between people and locations in huge databases of call records, email correspondence, money transfers and so on and so forth.

No.