DCGS-A versus Palantir: what the courts actually decide...

In 2011 and 2012, soldiers in Afghanistan were filing paperwork asking for a commercial software product by name. The Army had already spent years and billions of dollars building its own intelligence system to do the same job. A few years later, two federal courts told the Army it had broken the law in the way it tried to keep building that system. What the courts did not say is the part most people get wrong.

This is a procurement story, not a technology contest. It is about a specific legal duty the government owes taxpayers: when a commercial product might meet a need, the government has to genuinely check before it pays to build a custom one from scratch. The Army skipped that step, a court caught it, and an appeals court agreed. Underneath the legal fight sits a harder policy question the courts deliberately refused to answer.

What DCGS-A is

The Distributed Common Ground System-Army, or DCGS-A, is the Army's program-of-record system for fusing intelligence. Its job is to pull many separate classified feeds into one analytic picture: signals intelligence, satellite and drone imagery, full-motion video from unmanned aircraft, and human-intelligence reports. According to the Government Accountability Office, DCGS-A Increment 1 was tracked as a Major Automated Information System, the category the government uses for its largest, most expensive software programs.

One point matters before any numbers appear. "DCGS" is a family of systems, one for each service: Army, Air Force, Navy, and Marine Corps. DCGS-A is only the Army variant. Reporting frequently blurs the family with the Army program, and the two have very different price tags. Keeping them separate is the single most important thing to get right about the cost of this story.

The friction in the field

Around 2011 and 2012, deployed units in Afghanistan reported that DCGS-A was slow, hard to use, and unreliable. Some commanders submitted operational-needs statements, the formal mechanism for asking for equipment from the field, requesting a commercial platform made by Palantir Technologies. Their stated reason was that Palantir helped them analyze networks and pinpoint improvised explosive devices, the roadside bombs that were the signature threat of that war.

The complaints were not only anecdotal. In August 2011, the Army's Test and Evaluation Command assessed the baseline DCGS-A system and found it "not operationally effective." The same testing found the system rebooting on average roughly every eight hours. NPR reported in 2012 on the internal battle between soldiers who preferred Palantir and an Army invested in its own program of record.

Here the record demands precision. Soldier preference for Palantir is a documented complaint and a real morale issue. It is not an official finding that Palantir was technically superior across every intelligence mission the Army has to perform. Units reported and requested. That is a different thing from a verdict that one product beats another for the whole enterprise, and the distinction becomes the entire point once the courts get involved.

The 2012 report controversy

In July 2012, The Washington Times reported that an April 25, 2012 assessment by the Army's Operational Test Command had praised Palantir as easy to use while criticizing DCGS-A as too slow to process data. According to that reporting, a subsequent memo ordered copies of the April report destroyed, and it was replaced by a later version. Col. Joseph Martin, who headed the Army test command, signed a June 29, 2012 memo superseding the April report.

This episode is usually told as a cover-up. The record does not support stating it that flatly. The "destroyed report" framing rests mainly on Washington Times reporting and on Palantir advocates, including one member of Congress discussed below. The honest description is a controversy over a superseded or withdrawn report. It belongs in the timeline as an attributed dispute, not as an established finding that the Army buried inconvenient evidence.

Also in July 2012, Rep. Duncan D. Hunter, a Republican from California, publicly complained about Defense Department obstacles to wider field use of Palantir, amplifying the requests coming from deployed units. Hunter was a vocal Palantir advocate, and his congressional district neighbored the company's California base. That advocacy context is part of the record too.

The lawsuit

The dispute moved from field paperwork and press coverage into federal court in 2016, and the venue matters.

On June 30, 2016, Palantir USG, Inc. filed a pre-award bid protest in the U.S. Court of Federal Claims, docketed as No. 1:16-cv-00784. The complaint alleged that the Army's plan to award the next phase of the program, DCGS-A Increment 2, through a single-source development solicitation violated the Federal Acquisition Streamlining Act. That statute, codified at 10 U.S.C. Section 2377, requires the government to determine whether commercial items can meet its needs before it commissions a custom build. The company argued the Army had never seriously made that determination.

The plaintiff's full name was Palantir USG, Inc., and the target was the Increment 2 procurement specifically, not the entire DCGS-A program.

Judge Marian Blank Horn of the Court of Federal Claims ruled for Palantir. On October 31, 2016, she permanently enjoined the Army from awarding the DCGS-A Increment 2 contract until it complied with the commercial-items requirement. The merits opinion is reported at 128 Fed. Cl. 21 (2016); the opinion granting relief and the injunction is at 129 Fed. Cl. 218 (2016).

This is the event people mean when they say "Palantir won." It happened at the trial court, the Court of Federal Claims, in 2016. It was not an appeals-court ruling, and it was not a Federal Circuit ruling. Getting that label right prevents a common error.

The appeal came later. The U.S. Court of Appeals for the Federal Circuit affirmed the trial court in September 2018, in Palantir USG, Inc. v. United States, 904 F.3d 980 (Fed. Cir. 2018), No. 2017-1465. The appeals court held that the Army violated the Federal Acquisition Streamlining Act by failing to properly determine whether commercial items could meet its needs. So the arc is two dated events, not one: a 2016 win at the Court of Federal Claims, affirmed on appeal by the Federal Circuit in 2018.

What the ruling did and did not say

This is the load-bearing distinction in the whole story, so it is worth quoting the appeals court directly. The Federal Circuit expressly limited its holding to the procurement process. In its words, "we are not suggesting that the Army must choose Palantir as the awardee. We simply affirm that the Army must satisfy the requirements of 10 U.S.C. Section 2377."

The courts made no finding that Palantir's technology was superior. They found that the Army broke a procurement law by not doing proper market research and not first considering whether a commercial product could do the job. Those are different conclusions. One is about capability and is genuinely contested. The other is about legal process and is settled. The courts decided only the second one.

Stated plainly: the government lost because of how it bought, not because a judge decided its system was worse. Anyone who summarizes the case as "the courts ruled Palantir is better" is misreading it.

The money, with scope and year attached

Cost is where this story is easiest to get wrong, because at least three very different numbers float around and get merged.

The whole DCGS family, across all four services, has been reported to cost at least about $10.6 billion over its lifetime, with more than half, about $6 billion, already spent. This figure is attributed to GAO and repeated widely in press, but it could not be pinned to a single primary GAO document and year in the research for this piece, so it is reported as attributed to GAO rather than as a hard primary number. Two things about it are firm: it covers the whole family, not the Army alone, and it is a split between roughly $6 billion already spent and a roughly $10.6 billion or higher projected lifetime estimate. Spent and projected are not the same, and this figure is not the Army's bill.
Route Fifty reported in 2015 that, across the military, DCGS was estimated to eventually cost more than $10 billion. "Eventually cost" is a projected lifetime estimate for the family, not money already out the door.
The Army-only DCGS-A number is smaller and much softer. A commonly cited figure is around $2.3 billion for the Army system integrating hundreds of data sources, but that figure traces to secondary sources and varies depending on whether one counts a single increment, research and development only, or the full life cycle. Treat it as illustrative, not as an audited total, and never merge it with the joint-family $10.6 billion.

The one rule that keeps all of this honest: attach scope and year to every number, and keep money spent distinct from money projected. The central cost landmine in this story is folding a smaller Army-only figure into the larger whole-family total, or presenting a projected lifetime estimate as cash already spent.

The aftermath

After the courts forced the Army to consider commercial options, Palantir went on to compete for and win DCGS-A work. In 2018 the company was reported to have won a DCGS-A award valued at roughly $876 million. The system it had sued to be allowed to bid on became one it helped build.

There is a longer money thread worth naming without overstating it. Palantir received early backing from In-Q-Tel, the CIA's venture-capital arm, which is part of how public money seeds private intelligence technology. That is a sourced connection relevant to the broader public-money picture. It is not evidence of a conspiracy behind the DCGS-A fight, and it should not be read as one.

The honest failure critique and the honest mission defense

The critique. The Army spent years and billions of dollars building and defending an in-house intelligence system that its own deployed users repeatedly criticized as slow, unreliable, and hard to use. The Army's own testers called the baseline "not operationally effective," with reboots roughly every eight hours. A commercially available alternative was requested from the field through formal channels. Rather than fully evaluate that commercial option, the Army moved toward a single-source development solicitation for the next increment, and two federal courts found it had violated a law whose entire purpose is to make the government check for commercial solutions before it pays to build custom. Taxpayers funded a bespoke system whose users pushed back, and the government resisted an off-the-shelf option until a court ordered it to follow the commercial-items law. The 2012 controversy over a favorable field report being superseded adds a transparency question on top.

The defense. DCGS-A existed for a real reason. Modern Army intelligence has to fuse many separate classified sources, signals intelligence, satellite and drone imagery, full-motion video, and human-intelligence reports, into one picture, across security enclaves, at very large scale. The Army argued that no single commercial product at the time integrated all of those feeds and specialized functions, using electronic-intercept and SIGINT integration and drone-footage analysis as examples, and that a government-owned program of record with a single system integrator offered better long-term control, interoperability across the entire force, and projected multi-billion-dollar life-cycle cost avoidance. That cost-avoidance claim, sometimes cited as more than $3 billion, is an Army projection, not an audited result. Palantir was a powerful analysis layer for certain problems, notably IED and network analysis, but "easier for one deployed unit's mission" is not the same as "covers the entire Army enterprise intelligence requirement." The legitimate policy question was never whether Palantir was good software, because for many users it clearly was. The question was whether the Army must own an integrated backbone, buy commercial, or do both.

The courts did not resolve that policy question on the merits. They required only that the Army genuinely evaluate the commercial option first. And in the years after, DCGS-A itself moved to incorporate more commercial technology, which is arguably the oversight system and the procurement law working as designed.

Fact-check notes and sources

DCGS-A is the Army's program-of-record intelligence data-fusion system and DCGS-A Increment 1 was tracked as a Major Automated Information System: U.S. Government Accountability Office, GAO-15-282 (2015). DCGS-A is the Army variant of a service-specific DCGS family; the two scopes are distinct.
The whole DCGS family estimated at about $10.6 billion with about $6 billion spent, attributed to GAO: reported via press citing GAO, including Wikipedia's Distributed Common Ground System entry. This is a whole-family, projected-versus-spent figure that could not be pinned to a single primary GAO document in this research; it is reported as attributed to GAO and not applied to the Army program alone.
DCGS estimated to eventually cost more than $10 billion across the military (projected lifetime, whole family): Route Fifty (2015).
August 2011 Test and Evaluation Command assessment found the baseline "not operationally effective," rebooting roughly every eight hours; deployed units in Afghanistan requested Palantir: The Washington Times (Aug 7, 2012) and NPR (Sept 12, 2012). Framed as user complaints and requests, not an official verdict of technical superiority.
The 2012 superseded or withdrawn report controversy, including the June 29, 2012 memo by Col. Joseph Martin: The Washington Times (Jul 22, 2012). Attributed as a controversy; not asserted here as an established cover-up.
Rep. Duncan D. Hunter's July 2012 public criticism of obstacles to field use of Palantir: summarized in Wikipedia's Distributed Common Ground System entry. Note the advocacy context.
Palantir USG, Inc. filed a pre-award bid protest on June 30, 2016, No. 1:16-cv-00784, alleging the single-source DCGS-A Increment 2 solicitation violated FASA (10 U.S.C. Section 2377): GovInfo / U.S. Court of Federal Claims docket.
Judge Marian Blank Horn of the Court of Federal Claims ruled for Palantir and permanently enjoined the Army on October 31, 2016 (128 Fed. Cl. 21; 129 Fed. Cl. 218): CourtListener opinion and contemporaneous coverage in Defense One (Oct 2016). This is the trial-court win, not a Federal Circuit ruling.
The Federal Circuit affirmed in September 2018 (904 F.3d 980, No. 2017-1465) on FASA commercial-items grounds and expressly stated it was not requiring the Army to choose Palantir: U.S. Court of Appeals for the Federal Circuit opinion (Sept 13, 2018). The affirmance is 2018, distinct from the 2016 trial-court win.
The Army's stated defense, including the argument that no single commercial product integrated all required feeds and projected multi-billion-dollar life-cycle cost avoidance: Route Fifty (2015). Presented as the Army's stated rationale; the cost-avoidance figure is a projection, not an audited outcome.

DCGS-A versus Palantir: what the courts actually decided about the Army's intelligence system