I spent a week pulling Pentagon AI procurement data from USASpending.gov. Found that BAH was sitting on a $6.4B AI contract ceiling while trading at ~$9B market cap. They announced a $720M acquisition this morning confirming the thesis.
I've been building a pipeline on [USASpending.gov](http://USASpending.gov) \- 30 million rows of federal contract data going back to 2020. I've written a few pieces on this data before, mostly about DOGE cuts and defense spending patterns. This one is different because the timing was quite fortuitous.
**What is the CDAO?**
The Chief Digital and Artificial Intelligence Office was created by the DoD in 2022. It's the institutional brain for the Pentagon's algorithmic future - whatever AI tools the US military uses to identify targets, coordinate operations, and process battlefield intelligence flows through this office. If you want to know who's actually winning the defense AI race, you can read the procurement data.
**What the data show**
I filtered [USASpending.gov](http://USASpending.gov) transactions to CDAO as the awarding office. Here's the 2025 federal action obligation breakdown:
Total CDAO obligations 2025: $312 million
Palantir: $252 million — 81% of the total
Booz Allen Hamilton: $39 million
Johns Hopkins Applied Physics Lab: $6.5 million
The division of labor is visible in the transaction descriptions. Palantir's line items are all software - Maven Smart System enterprise licenses, Data-as-a-Service platform, Army Vantage analytics. BAH's line item is "business application development support services - labor." Johns Hopkins is doing basic research. Palantir builds the AI. BAH makes it work. Johns Hopkins figures out what to build next.
**The Palantir number: $23.5 billion**
The $252 million is obligated cash. The ceiling if the government exercises every option year is $23.5 billion against $1.8 billion currently obligated.
Project Maven alone has nearly $4 billion in enterprise license ceiling across its line items. The Data-as-a-Service platform adds $2.6 billion. The Space C2 data platform adds $2.15 billion. Army Vantage has multiple line items totaling over $4 billion in ceiling. Palantir is too rich for my blood - $286 billion market cap against $1.4 billion 2025 true FCF.
**But BAH? Here's where it gets interesting...**
BAH has one transaction line for CDAO in 2025. It's called CDAO Technology Synchronization of Business Operations - TSYBO. The 2025 federal action obligation is $39.2 million. The potential total value of that single vehicle is $6,408,923,808.
BAH's current market cap after significant compression in 2026 is roughly $7.6 billion. The market is pricing BAH as a legacy government consultancy in secular decline. The contract data suggests that it may hold the implementation layer for the Pentagon's AI buildout.
**Some caveats**
The potential end date on the TSYBO contract shows 2025 in my data. I don't fully understand what that means. It could indicate a re-compete requirement - BAH would need to defend its position as the incumbent. It could be a data artifact from how [USASpending.gov](http://USASpending.gov) records IDIQ vehicles. BAH has a strong historical renewal rate on major contracts but this is worth investigating before sizing up a position.
Also worth noting: this data is messier than my usual SEC XBRL work.If anyone here has domain knowledge on how IDIQ potential end dates work I'd love to hear it.
**BAH acquisition**
Today BAH announced a $720 million acquisition of Ultra I&C Mission Solutions, a defense tech company specializing in mission-critical software, encryption, and edge-compute products for battlefield deployment.
Think about what edge-compute and encryption hardware does in the context of a $6.4 billion AI integration contract. Palantir's Maven Smart System identifies targets and coordinates operations at the software layer. But that software has to run somewhere secure, at the edge of the network, in environments where connectivity is unreliable and security is paramount.
Full piece with all the data tables: [https://cavemanscreener.substack.com/p/bridges-to-nowhere-part-iii-inside](https://cavemanscreener.substack.com/p/bridges-to-nowhere-part-iii-inside)