Inside Palantir’s Army Business: From Bid Protest to a $10 Billion Ceiling

Calling the agreement a $10 billion Army contract compresses a decade of legal, product, and account-development work into the wrong financial unit. The July 2025 enterprise agreement gives the Army and other Department of Defense agencies a way to place orders for as long as ten years, subject to a maximum. The Army’s own announcement says the agreement makes no commitment to a specific purchase volume. [11]

That is not a semantic technicality. A ceiling measures possible buying capacity. An obligation is a legal commitment of government funds. Backlog depends on the issuer’s definition. “Remaining deal value” is Palantir’s company-defined measure, while remaining performance obligations are noncancelable contracted revenue not yet recognized. Revenue follows contract accounting. Those figures can describe the same customer relationship without being interchangeable. [12] [14]

Palantir’s Army history matters because it shows how a software company can turn access to a procurement into a larger account, without the dispute itself guaranteeing a sale.

The business choice came first

Palantir framed its business in its 2020 registration statement around reusable software platforms rather than one-off custom applications. Its account model had three named phases: Acquire, Expand, and Scale. The company described initial deployments at no or low cost, continued investment in accounts with negative contribution margins during expansion, and improving margins after the product became established across a customer.

[7]

That model is important to the Army story. A procurement written around a single developer and integrator can favor a labor-intensive build even when a commercial product might satisfy part of the requirement. A product company needs the customer to evaluate what already exists, what can be modified, and what truly requires new development. Otherwise, the vendor may be excluded before its product is compared on the merits.

Palantir already held defense work when it made that argument. In May 2016, U.S. Special Operations Command announced a firm-fixed-price license-and-support contract with a stated potential value of about $222.1 million. The notice said $5,002,225 was provided at award. [4] The contract notice predates the Army injunction. It also offers an early example of why face value and initial funding must be reported separately.

The procurement wall

The Army’s December 2015 solicitation for the second increment of the Distributed Common Ground System–Army sought one contractor to act as system architect, developer, and integrator. Palantir argued that the Army had not adequately determined whether commercially available technology could meet, or be modified to meet, the need.

The company did not simply “win its protest.” The Government Accountability Office denied Palantir’s pre-award protest in May 2016. [1] Palantir then filed a separate case in the Court of Federal Claims. In November 2016, that court found the Army’s commercial-availability analysis deficient and barred an award under the challenged solicitation until the Army complied with the statute.

[2]

The Federal Circuit affirmed in September 2018. Its limiting language is central to an accurate business history: the decision required the Army to perform the statutory analysis, but did not require the Army to choose Palantir or find that Palantir’s product met every requirement. [3] The legal victory changed the doorway. It did not choose the vendor.

The procurement form changes

The follow-on competition used a materially different form: multiple fixed-price indefinite-delivery, indefinite-quantity awards for a commercial combined hardware-and-software solution. In March 2018, the Army awarded positions to Palantir and Raytheon on a vehicle with a shared $876 million ceiling. The award notice said funding would be determined with each order.

[5] [6]

That record supports three restrained conclusions. First, a commercial-product path existed. Second, Palantir earned a position on it. Third, the shared ceiling did not hand either contractor $876 million. The customer still had to issue orders and obligate funds.

The distinction is useful beyond this one procurement. Winning the argument that a product should be considered is not the same as winning an award. Winning a place on a multiple-award vehicle is not the same as winning every order. And winning an order is not the same as recognizing its full value as revenue on the award date.

How access became expansion

Palantir’s filing offers company-reported evidence of account expansion, with important limits. It said Army accounts generated $51.9 million from 2008 through September 2018 and $134.5 million from October 2018 through June 2020. [7] The periods are unequal, the figures come from Palantir, and the comparison does not isolate litigation as the sole cause. It nevertheless shows that the post-ruling period coincided with a much larger Army business.

Later awards broadened the relationship:

• In March 2024, the Army announced a $178.4 million other-transaction agreement for ten Tactical Intelligence Targeting Access Node prototypes after competitive prototyping. [8] A prototype award does not establish production, delivery, acceptance, or fielding.
• In May 2024, the department announced a firm-fixed-price Maven Smart System contract with a stated $480 million value, with funding and work determined through orders. [9]
• In May 2025, it announced a $795 million modification for Maven software licenses. [10] A modification value does not automatically equal an obligation, cash receipt, or revenue recognized that day.

The records show an award position followed by higher company-reported Army-account revenue and broader awards. They do not isolate any one cause. Palantir’s own S-1 attributes government growth to platform improvements, larger contracts, and changes in government procurement. That multicausal explanation is consistent with the company’s Acquire, Expand, and Scale model. [7]

Five numbers readers should not combine

The public record now contains at least five Palantir government figures that answer different questions.

Figure	Date and scope	What it means	What it does not establish
Up to $10.0B	Army enterprise agreement, July 2025	Maximum potential value of purchases Army and other DoD agencies may place over ten years	Obligation, guaranteed orders, backlog, or revenue
$4.4B	Government remaining deal value, Dec. 31, 2025	Palantir metric that assumes available options are exercised and contracts are not terminated	Noncancelable government backlog
$12.3B	Aggregate government IDIQ value excluded from remaining deal value, Dec. 31, 2025	Awarded vehicles for which funding was undetermined or not guaranteed	A vehicle-level reconciliation or guaranteed sales
$4.9B	Company-wide remaining performance obligations, June 30, 2026	Noncancelable contracted revenue not yet recognized across government and commercial customers	Government-only or Army-only backlog
$1.848B	Government revenue, first half of 2026	GAAP revenue recognized from all government customers during six months	Army revenue, obligations, cash receipts, or profit

Palantir’s 2025 annual report is unusually helpful here. It excluded $12.3 billion of aggregate government IDIQ value from remaining deal value because the funding was undetermined or not guaranteed. [12] The filing does not provide a vehicle-by-vehicle bridge, so it would be speculation to assign a precise portion of that total to the Army enterprise agreement.

What the latest filing actually shows

Palantir reported $2.402 billion of government revenue for 2025. [12] For the first half of 2026, it reported $1.848 billion, an increase of 78 percent from the comparable prior-year period. In the second quarter, U.S. government revenue was $809 million and total government revenue was $990 million. [14]

Those figures show reported government revenue growth, but they do not disclose Army revenue. The government customer category includes customers outside the Army, outside the Department of Defense, and outside the United States. The quarterly filing also said that $428 million of the $437 million year-over-year increase in quarterly government revenue came from customers already present at the end of 2025. That supports an account-expansion reading at the government-customer level, not an Army-specific causal claim.

Revenue also says nothing by itself about program outcomes. The award notices cited here do not prove that every prototype was accepted, every option exercised, or every system fielded. Those require separate records.

Embedded does not mean exclusive

It would be easy to turn Palantir’s growth into an overbroad “control of the Army software market” thesis. The June 2026 Next Generation Command and Control announcement is a useful counterweight. The Army identified Anduril as the lead for a common-data initiative and described Palantir as a partner providing Foundry-related data infrastructure. [13]

That record establishes an important partner role, not the lead role in this initiative. A supplier can provide an important software layer without leading every acquisition.

The durable lesson for defense founders

Palantir’s Army history is not a simple courtroom-to-contract fairy tale. It is a sequence of business conversions:

1. Product architecture into a procurement argument. The company argued that the Army had to evaluate available commercial software before defaulting to a custom build.
2. A legal opening into a competition. The injunction required a proper analysis; it did not award Palantir anything.
3. A vehicle into funded work. A place on an IDIQ created access, while orders and obligations created actual government commitments.
4. A foothold into account expansion. Palantir reported higher Army-account revenue, while Army records document additional programs and an enterprise purchasing mechanism.
5. Contract activity into accounting results. Contract ceilings and GAAP revenue must be tracked separately.

That sequence is why contract vocabulary matters. The wrong label hides the strategy. The right one shows how a software supplier can move from procurement access to orders and recognized revenue over time.