Palantir S-1
sec.gov
sec.gov
"The engineering elite of Silicon Valley may know more than most about building software. But they do not know more about how society should be organized or what justice requires.
Our company was founded in Silicon Valley. But we seem to share fewer and fewer of the technology sector’s values and commitments.
From the start, we have repeatedly turned down opportunities to sell, collect, or mine data. Other technology companies, including some of the largest in the world, have built their entire businesses on doing just that.
Software projects with our nation’s defense and intelligence agencies, whose missions are to keep us safe, have become controversial, while companies built on advertising dollars are commonplace. For many consumer internet companies, our thoughts and inclinations, behaviors and browsing habits, are the product for sale. The slogans and marketing of many of the Valley’s largest technology firms attempt to obscure this simple fact."
Oh well ok then, they don't do any data collection or mining, they just sell the tools to empower three-letter agencies to do it more effectively. Conscience is clear then!
I imagine that before Palantir came around, the analysts at these agencies did what most other analysts anywhere do and used Excel to create sophisticated pivot tables when working on {insert defense operation}.
Where does it stop being okay? To the extent that it even matters, Palantir appears to at least own up to the fact that they are used by governments.
> socially useful purpose
Who decides what's "socially useful"?
Even per their S-1, their market sizing shows "Commercial" to be a substantial proportion of their aspirations:
> Commercial. Our estimate of the TAM in the commercial sector is $56 billion.
> Government. We estimate the TAM in the government sector, including government agencies in the United States, its allies, and in other countries abroad whose values align with liberal democracies, is $63 billion.
> The company’s top three customers — which aren’t disclosed — together represented 28% of the company’s revenue for 2019. Its top twenty customers represented 67% of total revenues, with each one of those customers averaging $24.8 million in revenue.
They admit they count every single government sub-agency as a different customer (their example is NIH and the CDC would count separately despite the same parent organization) so the top 25 is basically just governments. There is incredible revenue concentration there.
Also, it’s important to consider their TAM projections because 1) those are the only numbers we have in this S-1 that specifically delineates government vs non-government, and 2) their aspirational valuation is predicated on this TAM. It’s who they think they will grow up to become.
Their largest project is Palantir Gotham which is exclusively for government agencies (federal, state, ICE, and by Project Maven after the tech companies abandoned it due to ethics concerns!)
You’re talking primarily about Palantir Metropolis which is used by hedge funds. Not exactly some beneficial use of technology to counterweight working with ICE to separate children from their families.
They don’t list their top customers because they are top secret government agencies but if you look at their customers then it’s clear that the ones paying them the most are secretive government agencies and it’s been that way since day 1
Did you even read their S-1? Palantir Gotham is NOT “exclusively for government agencies”. According to their own filing, it’s used by their Commercial team too. It even says so on their website.
> You’re talking primarily about Palantir Metropolis which is used by hedge funds. Not exactly some beneficial use of technology to counterweight working with ICE to separate children from their families.
If you’re going to attack them, do it honestly by accurately representing them. Palantir isn’t used just “used by hedge funds”, if you actually read their prospectus and their website, their use cases include: auto racing, automotive parts manufacturing improvement, corporate cyber security, regulatory compliance (incl GDPR), pharma/drug development, and airline route optimization.
https://www.palantir.com/solutions/
> but if you look at their customers then it’s clear that the ones paying them the most are secretive government agencies ande it’s been that way since day 1
Yeah this is conjecture, at best. Nobody disputes that Palantir works with the government, but you don’t have enough information to claim with certainty that they work only with the government, and how much of their revenue comes from the 3-letter agencies. Unless you can actually substantiate your claims, these are just unfalsifiable hunches.
1. "We generated $742.6 million in revenue in 2019"
2. "The total remaining deal value of the contracts that we had been awarded by government agencies in the United States and allied countries around the world, including existing contractual obligations and contractual options available to those government agencies, was $1.2 billion"
3. "We do not include government contracts totaling $2.6 billion, as of June 30, 2020, that we have been awarded where the funding of such contracts — also known as indefinite delivery, indefinite quantity (“IDIQ”) contracts — has not yet been determined."
So they have 1.62 years of committed revenue from governments and 5.12 years of revenue including the IDIQ contracts. Even assuming the contracts are spread over five years, which seems to be their long term contract length, it indicates their revenue is overwhelmingly from the government.
-----
As for Palantir Gotham being for government agencies, here is the language from their S-1:
1. "We built Gotham, our first platform, for government operatives in the defense and intelligence sectors...Gotham is now used broadly across government functions."
2. "Gotham enables users to identify patterns hidden deep within datasets, ranging from signals intelligence sources to reports from confidential informants, and helps U.S. and allied military personnel find what they are looking for. We later found that the challenges faced by commercial institutions when it came to working with data were fundamentally similar. Companies routinely struggle to manage let alone make sense of the data involved in large projects. Foundry was built for them."
See how it says Foundry was built for commercial and Gotham for "government agencies"?
Let me just pull the relevant quote from the S-1 directly for you:
> In 2019, 53% of our revenue came from commercial customers and 47% came from government agencies.
It's on page 88, in case you didn't make it that far.
Metropolis got deprecated (read renamed) Foundry because they kept screwing up their sales and decided a rebrand would be more effective on their commercial side.
https://www.quora.com/What-are-the-main-differences-between-...
per someone who runs tech on the commercial side,
Foundry started as our attempt to create scalable process and rigor around data integration. There's a longer story to tell here, but the short version is that we started out with off-the-shelf orchestration systems for running jobs on a schedule (think Jenkins and Rundeck — things that were more robust than cron). At some level, that worked, and we had to figure out how to scale it, leading to a combination of HDFS, Rundeck/Jenkins, a git repo, and a common language for mutating data.
But since you asked, here is an old article about it: https://www.bizjournals.com/sanjose/news/2014/09/15/cia-back...
Title: CIA-backed Palantir adds $50M to its war chest
“Palantir's backers include the CIA's In-Q-Tel venture fund and Thiel's Founders Fund. Its customers include government intelligence agencies such as the National Security Agency.”
Google In-Q-Tel and Palantir and you’ll find 100 other articles about it.
For better or for worse, the American electorate approves of the CIA/FBI -> https://news.gallup.com/poll/224804/republicans-push-governm.... This is even more the case for defense, which is just about the only American institution that has had a reliably consistent approval rating -> https://truthout.org/articles/why-does-the-us-military-have-...
Alex Karp's CEO letter makes a similar point: "The engineering elite of Silicon Valley may know more than most about building software. But they do not know more about how society should be organized or what justice requires."
No one and everyone decides what is socially useful. But it's fairly uncontroversial that invasion of privacy is not socially useful.
The majority of Americans appear to approve of the CIA/FBI -> https://news.gallup.com/poll/224804/republicans-push-governm...
“Socially useful” is well and good, right up until you disagree with the masses what constitutes “socially useful”. Silicon Valley engineers have input into that decision, but don’t get the final say.
Also, the NSA and CIA are act unconstitutionally and then hide the fact they are doing so from the voters. A fairer way to see if there is actually public support would be to allow a group of 100 people into the CIA/NSA archives and allow them full access to the entirety of the uncensored information, without anyone to justify what they have done beyond what has been recorded.
Otherwise, polling people on what unaccountable agencies do illegally in secret is (figuratively) like asking a North Korean if Kim Jong Un is doing a good job. They just don't have the information to judge.
This is literally the premise of clandestine agencies, and it’s not a new concept. They’ve been around for 50 years now. This isn’t some bait-and-switch, voters have always known what a clandestine agency is.
I understand that there exists some voters who don’t believe that clandestine agencies should exists. Unfortunately, such voters are in the minority. And unfortunately the majority of Americans find clandestine agencies (and in general, a strong defense) to be “socially useful”.
Again, I am directly refuting that argument. Voters can approve of clandestine agencies if they sign onto the fact that there exists some organs of the government that engage in secret activities unbeknownst to the public. What you are saying would only be correct if at first voters expected to be privy to the activities of the clandestine agencies, and then that happened to change. That has not been the case.
I have provided you an actual poll that asks users "do you approve of these secret agencies (that have always been secret)". When presented with evidence that voters do approve of that, it's a bit of a hack to say that it's literally impossible for voters to approve of it. Not every voter in America is a progressive programmer that subscribes to the EFF's world view.
So no, just being polled highly is not sufficient for something being democratically sanctioned. If you want to know if the People agree with what the CIA does let 100 people read the entire archives uncensored without any further justification and see what you get.
Quite the gulp of koolaid to swallow straight faced
Indeed, that definition is definitely how I, and most tech-oriented folks I interact with, would define data mining. First and foremost.
But as stated in a different comment thread above, they don't "do" data mining, they just sell data mining tools and how-to manuals! Totally not the same thing!
It's one thing for consumer internet companies to show me an ad for mattresses because they know I searched for mattresses. It's wholly another to help government entities aggregate license plate data, run "predictive policing" programs, run the Pentagon's spy drone program, and help ICE track immigrants, as Palantir does. Frankly if given the choice I'll take the mattress ads.
they'll collect or mine whatever they can get their hands, i'm certain of that. it's a data company.
Palantir produces data analysis tools, this is basically Excel but for bigger data sets. Microsoft doesn't get what you put into your Excel sheets (not really sure now...) and neither does Palantir with their platform.
They probably have some sort of telemetry to monitor how their tools are used, to "improve user experience", like most large software vendors. But it is not related to the data being processed.
Are you familiar with Project Maven? They gather drone data and have humans label it so they can plan drone strikes and other intelligence efforts.
The only real solution I've found is to help build the fully agent-centric web, where there are no third parties - full P2P Protocol Cooperativism. [1] [2]
[1] https://medium.com/holochain/holochain-reinventing-applicati...
Palantir the company doesn't have access to the data. The government agencies have access to the data. Palantir licenses its software to these organizations. The company doesn't have special access to any sensitive data.
Well that's a good reason.
The point I'm making is that I don't agree on Palantir's approach and strategy. If you watch the Ava Duvernay's documentary '13th' you'll see that systemic oppression of black bodies never ended. The school-to-prison pipeline guarantees exploitation and incarceration of the US's dark history of slavery and genocide.
So instead of Palantir pushing for systemic change, I see their technology as regressive and focused on giving power to unjust causes.
If you want to read more about how algorithms recreate bias, there's quite a lot of material that has come out recently. Here is an important point from Algorithsm of Oppression:
※ Those who have the power to design systems--classification or technical [like library, museum, & information professionals]--hold the ability to prioritize hierarchical schemes that privilege certain types of information over others.[1]
There's also another book called Automating Inequality, among others.
[1] https://www.goodreads.com/review/show/2206336831?book_show_a...
There seems to me to be a strong neoliberal/libertarian population on Hacker News who believes, as Margaret Thatcher famously said, that 'there is no alternative'. They have completely soaked up Silicon Valley's Imperialist 'Manifest Destiny' mission.
They can't see their own privileged position and power. They attribute their own success to their personal efforts, instead of seeing how various social institutions (including knowledge and access to a rich scientific inheritance) built them up.
So when we pose them a question about Project Maven, they want to identify the one 'bad apple' [1] instead of facing the fact that systemically something is rotten, and that we might have to evolve the underlying architectures. [2]
[1] https://www.youtube.com/watch?v=KMNZXV7jOG0&feature=youtu.be
[2] https://medium.com/holochain/holochain-reinventing-applicati...
It's clear that the authors of the US Constitution believed that it should be difficult for the government to perform certain enforcement actions. The last 200+ years of jurisprudence and legislation back this up. The police need a warrant in order to search a home. The prosecution needs a court order to compel testimony.
I don't have a problem with saying "the LAPD is going to enforce traffic laws whether or not Palantir assists them". But my response would be "yes, and we should think very carefully about how easy we make it for 'enforce traffic law' to look like 'mass surveillance'". Legislation and jurisprudence move a lot slower than we'd like and I think it's good that individuals and companies think about how they choose to participate in these systems and if they're over-enabling government enforcement and surveillance while we wait for politicians and judges to play catch-up.
this isn't true actually. the police uses services like Palantir as a sort of get around because lots of data collection literally requires a warrant[1]
The US suffers from a schizophrenic legal situation where it has (for good reasons) barred its government from spying on its citizens, but has somehow outsourced it to private entities, the one level of indirection apparently makes that okay.
[1]https://arstechnica.com/tech-policy/2020/07/cbp-does-end-run...
This is a pattern I've seen from America in general, loopholes against the spirit of the law, healthcare through 3rd parties, it seems there is always somewhere to kick the buck down for liability purposes. There is always a quest to find a technicality for a way out of some rules, the spirit of the rules don't seem to matter much.
I'd say they're still a Silicon Valley company at heart.
Too bad the only government department concerned with that is the Department of Energy, and last I checked Palantir isn't a contract partner running any of the national labs.
They sure do fund cool basic science research though.
They are both bad and Palantir's wrongs are, if anything, far more apparent https://theintercept.com/2017/03/02/palantir-provides-the-en...
https://techcrunch.com/2020/08/20/palantir-denver-hq-ipo-dir...
How useful would this software be without data from other companies?
-- 2019 Financials ('000s)
Revenue: 742,555
Cost of Revenue: 242,373
Gross Profit: 500,182
Sales and Marketing: 450,120
-- Sales and Marketing includes pilot costs
"Sales and marketing costs primarily include salaries, stock-based compensation expense, and benefits for personnel involved in executing on pilots and performing other brand building activities, as well as third-party cloud hosting services for our pilots, marketing and sales event-related costs, and allocated overhead. The Company generally charges all such costs to sales and marketing expense in the period incurred."
but given Covid maybe they’re saving a bunch of money on employee travel
- 86% gross margins (122,725 / 142,606)
- 86% YoY rev growth (76,770 to 142,606)
- 75k paying customers (easy come, easy go)
- one-off sale pitches vs one-off lines of code
https://www.sec.gov/Archives/edgar/data/1477720/000119312520...
Asana and Palantir both have terrible financials. They've both basically doubled their net loss YoY.
That's cuz they're not. They are a hybrid services business that also sells "cloud computing". The are very comparable to Cloudera - they have a proprietary analytics platform that they then customize specifically for the workload. They then charge an implementation fees, on-going "maintenance", and a subscription fee for the "software" (which is basically a highly customized single tenant).
> Gross margin of 67%
Which is still better than IBM or any professional services firm could ever achieve from the government.
> If you add these pilot costs to COR, you'd get better numbers manufacturing furniture.
If you added S&M costs to revenue attributed to a multi-year subscription fee, you'd be called an idiot. Hence why they are probably not in there (could be wrong...we don't know for sure).
I do not like Palantir in the slightest, but this assessment is kinda off.
IBM, on the other hand, has a 2x revenue multiplier.
Maybe this isn't the best metric to use, and Palantir does have better margins, but I just can't see these numbers being worth their current valuation as a services business.
You're right. Discounted Cash Flow is.
California taxes capital gains at 13%, and 9 states have none (Alaska, Florida, Nevada, New Hampshire, South Dakota, Tennessee, Texas, Washington, and Wyoming). As an employee, you could save $130,000 on your $1 million dollar gain.
That pays for an awesome rental house in Seattle or the Nevada side of Lake Tahoe. It will be a lot more tempting since you can just ride out covid for a year or so and keep working remotely like you are now. You can time this to your liking since the trigger is you selling the stock, so you can move in 2021 and sell in 2022. Realistically having 1 successful IPO is very lucky, maximizing the return makes a lot of sense.
And I think this is much a much sharper decision than for public company employees. You will have been accumulating stock for 4-10 years and it's been completely illiquid. You get this big hammer of a tax bill when you sell, and it's very tempting to optimize it. This is always happened a little but in covid times, it seems like it will be 10x of the usual. Expect a lot of complaints from Seattle and Austin :-).
(added attempt to)
If you worked in California while you were earning that stock / options, they'll come after you for the taxes.
California knows. They know that people want to try and have tried to dodge these taxes. They will find you and get their pound of flesh.
I admit I wouldn't know how options would play out here, but if the reason for the move is 'avoid taxes', they have probably done their due diligence.
RSUs with a double trigger on IPO will still be taxed on vesting as if you were in CA.
Anyway, making $1m gain as an employee isn't really a thing here either.
Our platforms contain “open source” software, and any failure to comply with the terms of one or more of these open source licenses could negatively affect our business.
Where as pivotal says that being an active part of a vibrant open source community is crucial:
If open-source software programmers, many of whom we do not employ, or our own internal programmers do not continue to use, contribute to and enhance the open-source technologies that we rely on, the market appeal of our offering may be reduced, which could harm our reputation, diminish our brand and result in decreased revenue.
Like this makes sense, palantir is a closed source first kinda company and obviously a different value to its customers than pivotal. Kinda interesting to see that manifestation in the risk factors though.
[1] https://www.sec.gov/Archives/edgar/data/1574135/000104746918...
- $2.2M in salary ($225k base + "stipend" + "parental benefit" + "salary adjustment")
- Bonus of $21k (?)
- $2.5M in stock
- $11.3M in options
- $25k in "other compensation)
All total $16M.
The COO got $26M.
This is all "compensation paid for the year ended Dec 31, 2019".
Doubly so when the core business driver is government connections.
So the $11M is the incremental increase in value of the new options versus the old options.
"The amount reflects the incremental increase in the fair value of the stock option to purchase 12,401,568 shares of our Class B common stock originally granted to Mr. Cohen in August 2015 with an exercise price of $7.05 per share arising from the cancellation of such stock option in exchange for the grant of a new stock option with an exercise price of $6.03 per share on November 6, 2019"
But what's your leverage to get it?
Obviously he had enough leverage to get it.
> The company’s top three customers — which aren’t disclosed — together represented 28% of the company’s revenue for 2019. Its top twenty customers represented 67% of total revenues, with each one of those customers averaging $24.8 million in revenue.
[0]: https://techcrunch.com/2020/08/21/leaked-palantir-s-1-shows-...
> Our top three customers together accounted for 33% and 28% of our revenue for the years ended December 31, 2018 and 2019, respectively, and 31% and 29% of our revenue for the six months ended June 30, 2019 and 2020, respectively.
> Palantir notes that customers from different parts of the same government department or company are considered separately (Palantir’s example is that the CDC and NIH are both part of the Department of Health and Human Services, but would be billed separately and are thus considered separate customers for the purpose of its calculation).
Likely all government entities with 5~10 year contracts. Customer concentration is only an issue when there is a possibility of churn. This isn't the case.
There's a "class F" of shares which are held only by Alex Karp, Stephen Cohen, and Peter Thiel, with each owning 1/3. Class F has "voting power representing up to approximately 49.999999%".
Does that mean that basically Karp, Cohen, and Thiel have full voting control?
> So long as our Founders [...] meet a minimum ownership threshold on the applicable record date for a vote of the stockholders[...], these Founders will effectively control all matters submitted to a vote of the stockholders for the foreseeable future. This could delay, defer, or prevent a change of control, merger, consolidation, or sale of all or substantially all of our assets that our other stockholders support. Conversely, this concentrated control could allow our Founders to consummate a transaction that our other stockholders do not support.
I’m not saying I want this. But in my eyes it is the inevitable next step to prevent law enforcement events that end in death while still being able to stop criminal activity sometimes before it happens.
Tech eats the world and old methods of policing are ripe for disruption.
Some snazzy CSS selectors + js magic + pdf.js something something perhaps? Because SEC doesn't provide nice PDF's, right? I haven't found anything at least.
> Financial statements in XBRL also are required as exhibits to Securities Act registration statements that contain financial statements, such as Form S-1 (except registration statements filed in connection with an initial public offering).
And by the looks of the search result pages for Apple [1] and Palantir [2], it looks like they haven't filed it that way (the "Interactive Data" label is missing):
[0]: https://www.sec.gov/rules/final/2018/33-10514.pdf [1]: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&C... [2]: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&C...
I suppose there must be a reason why we don't.
Because you're asking your portfolio manager to make you a custom index fund. The whole point of an index fund is to spread risk among lots of companies and sell fragments of that risk to lots of people. You can make your own custom index fund, it's called buying a share of every stock other than the ones you don't want.
You could short them to the same weighted percentage that they exist in the index but that's of course not completely simple and you have to pay fees for the shorted stock. If O&G had a convenient ETF (I don't know if there is or not, I imagine there is), you could short that rather than individual companies.
Shorting Facebook would be... interesting for your returns.
An infrastructure bill was badly needed in Q2/Q3 to have impact in time. As it stands now it will hit too late and another round of USGOB aid will be needed by companies small and large. This will happen regardless the extent of fall and winter quarantine needs.
The inability of Congress and the White House to put politics aside is going to cost American workers and companies dearly. It will not have been worth the political gains either side expects.
There was never a time in this Country's history that bipartisanship was this needed, at least since WWII. The failure defies words because even in less critical times names like Baker, Johnson, Dole, Rayburn, O'Neill, Rostenkowski and others - some corrupt - put self-interest aside briefly to conduct the country's business.
We are to blame by the way. The HN community is comprised of many bright, successful and/or wise folks from all political spectrums. We're either caught in the partisanship, or we're silent and checked out.
Wait, what did they do or know about organizing society ?
You can see how much prepaid money is "waiting" to be recognized by looking at Deferred Revenue on the balance sheet. $263 million at the end of 2019.
[1] https://www.investopedia.com/terms/r/revenuerecognition.asp
accrual accounting is the devil's work.
They have 125 customers. Even assuming everey single one of them came online last year, we're looking at millions of dollars each?
That's a lot of…bribes? Idk even know what could cost that much.
I don't think they'll ever be "stamping out software" with its associated margins like a SaaS or traditional software company.
EDIT:
- If this first ddg result is to be believed there are 20% more IPOs than last year https://stockanalysis.com/ipos/2020-list/
- There's actually a thread on this, so perhaps downvote this and read/comment/speculate there https://news.ycombinator.com/item?id=24265456
I've been really reminded of the Big Short, housing market shot up before the big crash so others would be left holding the bag... musical chairs isn't just for preschoolers I guess.
I think this theory tends to be based on the idea that the current state of the stock market does not reflect reality for many businesses (and the economy as a whole), nor their future prospects, and that a correction is imminent.
Obviously this is speculation, but it's interesting because someone probably does know - there's probably some overlap in the advisory banks for these IPOs, and they must have some idea of the thinking of the executive teams of these companies.
One explanation that was offered was that there are a ton of IPOs all the time: https://www.nasdaq.com/market-activity/ipos
These just made it on to HN as there were some big names in the mix (Asana, AirBnB, Unity, SumoLogic, Snowflake, Palantir, etc). Now, its not clear to me why all of these larger tech companies are going for it this week however...
Do you recognize even half of those? And this is just one exchange in one country.
So if there are more IPOs and more tech IPOs that's kinda weird on weird.
IPOs have become less common due to regulations as well as adequate growth of the private secondary market following more coordination by accredited investors, but they used to be 3x more common and at much lower valuations.
After the election, it probably will. No matter the result.
now we just have to wait and see what the “cost” of it all is because we’re all treating it as a free lunch
https://www.economist.com/business/2020/08/22/initial-public...
(Discussed recently: https://news.ycombinator.com/item?id=24232199)
TLDR, comparing the year-to-date, tech IPOs are on the high end historically, but not outlying the historical range (I make it just in the top quartile for IPO count). Considering all IPOs (tech and non-tech) though, there are more than any time in the last 15 years.
So yes there are more IPOs, but that doesn't explain why there are more tech IPOs being posted on HN.
- Palantir: https://www.sec.gov/Archives/edgar/data/1321655/000119312520...
- Cloudflare: https://www.sec.gov/Archives/edgar/data/1477333/000119312519...
- Zoom: https://www.sec.gov/Archives/edgar/data/1585521/000119312519...
[1] https://www.sec.gov/Archives/edgar/data/1809122/000110465920...
A draft is created, and then the company's executives, their legal, the accountants, investment bankers, and investment banker's legal all get in a room and go page by page through the document for several days. The legal team stays up late making the edits each night. Then it gets submitted to the SEC for review, and you get comments from them that have to be addressed. You submit the revised document with responses to their comments and keep doing that until the SEC has no comments.