How Palantir Built a $130 Million Data System for the IRS

The Internal Revenue Service has paid Palantir Technologies more than $130 million to build and operate a data analysis system that searches federal databases to support financial crime investigations, according to new reporting from The Intercept. The arrangement, spanning over a decade, represents one of the largest partnerships between a defense technology firm and a civilian tax agency.

The relationship began in 2014, when the IRS first bought technology from Palantir, a company founded by Peter Thiel, according to government contracting records reviewed by Wired. What started as targeted software tools has grown into a comprehensive system that now touches both criminal investigations and routine compliance work at the agency.

The Custom Analytics Platform

At the center of this partnership is a custom tool called SNAP—Palantir's Selection and Analytic Platform. It was designed specifically to help IRS officials decide which tax returns to audit and to identify patterns that might point to tax evasion.

For decades, the IRS used a statistical model called the Discriminant Information Function, or DIF score, to flag returns for review. SNAP replaces that older approach. Before the Palantir system arrived, the IRS was running more than 100 separate computer systems and 700 different methods—cobbled together over decades—to select cases for investigation. SNAP consolidates all of that into one platform.

The new system analyzes financial transaction flows and looks for connection patterns that could suggest illegal tax schemes. In 2023 alone, the IRS paid Palantir $1.8 million specifically to improve SNAP's ability to identify which cases to audit. Internal IRS documents refer to a "Discoverer Replacement/Palantir Solution," showing that SNAP has replaced older investigative tools.

How the System Fits Into the IRS

The IRS Criminal Investigation division now includes a formal "Advanced Analytics unit," part of what the agency calls its "Advanced Analytics and Innovation strategy." This infrastructure supports both routine investigations and proactive searches for suspicious patterns across multiple federal databases.

The system cross-checks financial transactions against taxpayer data held elsewhere in government and uses automated scoring to flag potential targets for enforcement. What these algorithms look for includes shell company networks, unusual payment patterns, and coordinated tax avoidance schemes that span multiple states or entities.

We have seen similar shifts before. When defense contractors moved from military applications into civilian government work in the years after 9/11, many of the same surveillance techniques developed for overseas intelligence operations were adapted for domestic use. Palantir's data consolidation approach now applies those capabilities to financial records at a scale that would have been difficult to imagine two decades ago.

The Broader Context

The timing of this expanded partnership matters. The IRS received billions in additional funding through the Inflation Reduction Act, which included money specifically for better enforcement tools and technology upgrades. This Palantir contract sits within that larger modernization push.

The platform processes what Palantir describes as "sensitive federal databases"—records of financial transactions, payment histories, and intelligence feeds from across government. Its machine learning algorithms can spot patterns in massive datasets that human investigators working alone would struggle to find. This could make the IRS more efficient at catching complex financial crimes.

But this capability also raises legitimate questions about the scope of financial surveillance that citizens face from their tax agency. When dozens of federal databases sit under a single analytical platform, the system creates an unprecedented view into individual financial behavior. How tightly that power is controlled, and who oversees its use, are questions worth asking. The IRS has been relatively quiet about how these algorithms work and what safeguards exist to prevent abuse.

Palantir's relationship with the Trump administration has also drawn attention. Founder Peter Thiel was a major donor to Trump's 2016 campaign and maintained ties to administration officials. The New York Times previously reported that Palantir pushed data-sharing initiatives across federal agencies during that period. Transparency advocates, including American Oversight, have filed lawsuits against multiple agencies, including the IRS, to obtain records about how Palantir systems are actually used.

What Comes Next

The $130 million contract covers multiple years of platform development, data integration, and ongoing support. This is among Palantir's largest civilian government contracts outside immigration enforcement, where the company has provided similar data-fusion capabilities to Immigration and Customs Enforcement.

The shift toward algorithmic case selection and automated pattern detection marks a significant change in how the IRS conducts enforcement. The system's ability to process vast datasets and spot subtle connections could increase efficiency, especially for complex financial crimes that older audit-selection methods would miss.

As the IRS continues modernizing, the Palantir partnership is likely to expand further. Additional data sources and new analytical functions beyond current tax enforcement could eventually be added. This reflects a broader trend: specialized defense contractors are increasingly providing data infrastructure for civilian government agencies. How that relationship unfolds—and whether robust oversight keeps pace—will shape the future of federal enforcement operations.