Former US military and intel officials express concerns about Palantir, arguing its success is largely down to its software having a friendlier UI than rivals
New York MagazineSharon Weinberger
Context & Ripple Effects
The UI critique lands on a company whose moat has always been contested. Wired's 2017 reporting traced how Palantir built an opaque surveillance network for police departments and agencies, locking clients in through deep data integration rather than polish. If former military and intelligence insiders are right that the real differentiator is a friendlier interface, that lock-in story looks thinner than it did.
The timing matters because the government business is already under pressure: the Wall Street Journal reported in early 2024 that Palantir's [[a:848530|US government revenue growth had slowed as agencies weighed cheaper rivals and pushed back on data-access limits]]. A reputation argument from inside the national-security community compounds the commercial risk that insiders flagged later over the company's enthusiastic embrace of Trump policies.
First-order effects
Rival government-data contractors gain a ready-made sales argument: match or approach Palantir's interface and the premium case for switching costs collapses.
Palantir's own pitch to agencies — that its platform is irreplaceable infrastructure — now has to survive testimony from the very officials who used comparable tools.
Second-order effects
Agencies already shopping for cheaper alternatives with fewer data-access restrictions have a lower barrier to exit if the differentiation is cosmetic, accelerating the revenue slowdown the Journal documented.
The critique pressures Palantir to defend value through capabilities beyond presentation, just as frontier AI labs intensify competition for the same enterprise and government workloads.
Third-order effects
If interface quality, not proprietary data plumbing, decides government software wins, procurement tilts toward vendors who package commodity capability well — eroding the deep-integration lock-in model Wired described in 2017.
A national-security customer base willing to say publicly that a flagship vendor is overrated raises the bar for any state-aligned AI company claiming mission-critical status.
The trend: Government AI and data-platform advantage is migrating from proprietary data integration toward interface and packaging, making incumbents' lock-in easier to challenge.
I've heard that Palantir's tech stack is outdated (even though they talk about using AWS and Azure in their S-1). This article also claims that the software is clunky, needs to be heavily customized, and relies on a lot of manual labor behind the scenes: https://nymag.com/...
The answer is ‘yes.’ Unproven outcomes, analytic capabilities that are now widely available from competitors - Palantir's main advantage is the firm's political connections and ability to avoid competitive bidding procedures https://twitter.com/...
“Thiel and Karp have effectively positioned Palantir as a pro-military arm of Silicon Valley, a culture dominated by tech gurus who view their work as paving the way for a global utopia.” - One or two other companies doing this but Palantir is definitely an OG. Great piece. https…
An inside look at how Palantir — Big Data's scariest, most secretive unicorn — got its start, and whether its proprietary surveillance technology is more smoke and mirrors than crystal ball. @weinbergersa reports https://nymag.com/...
First, everyone agreed they have a good user interface. That's not a small thing. They also agreed it took a lot of manpower to make Palantir software work. That is, in fact, what some people praised: the military loved Palantir's engineers who worked with them in Afghanistan.
“Palantir, it turns out, has run headlong into the problem plaguing many tech firms engaged in the quest for total information awareness: Real-world data is often too messy and complex for computers to translate without lots of help from humans.” https://nymag.com/...
I started off just interviewing people. Anyone I could find who had worked for Palantir, with Palantir, or in some cases, against Palantir. People have really strong opinions on the company - for and against - so I looked for the things everyone agreed on.
The question financial analysts often asked about Palantir: Is it selling software, or is it some hybrid of software and consulting? The latter, according to financial experts, is probably going to be a low margin business. Not something worth 20 times revenue.
So what's the problem? Nothing, unless you want to be valued at more than $20B with less than $1B in revenue, Those “forward deployed engineers” are really expensive.
Palantir goes public today, so I want to talk about my deep-dive for @NYMag on the company. In 20 years writing on military tech, I've found the most useful question to ask first is not if something is scary, but if it works as advertised. https://nymag.com/...