Defense Tech 101: How Palantir, Anduril, and the New Guard Are Changing Military Procurement

Executive Summary

U.S. defense procurement is undergoing the most significant structural shift since the post-Cold War consolidation of the 1990s. The traditional defense industrial base — Lockheed Martin, Raytheon/RTX, Northrop Grumman, General Dynamics, L3Harris — still captures 70%+ of prime contract dollars, but a cohort of software-first, commercially-oriented defense technology companies (Palantir, Anduril, Shield AI, Joby Aviation, Joby Defense) is growing contract value at 40–70% annually and changing how the Pentagon thinks about acquiring technology. Understanding how money actually flows through the defense budget, which procurement mechanisms the new guard is using, and where the investment opportunity is concentrated is essential for investors seeking exposure to the $900B U.S. defense budget.

How Defense Procurement Works (and Why It's Broken)

The Federal Acquisition Regulation (FAR) governs most defense procurement. The FAR's full contract process — requirements definition, RFP, proposal, source selection, award, protests, modifications — can take 2–5 years from initial concept to funded contract. For a fighter jet or aircraft carrier, this timeline is acceptable. For a software system or drone, it is catastrophic: the threat environment evolves faster than the procurement process.

The root structural problem is the "requirements creep" cycle:

1. Service (Army, Navy, USAF) defines requirements based on current threat assessment.
2. 18–24 months pass during requirements refinement and RFP drafting.
3. Prime contractors submit proposals based on the now-outdated requirements.
4. Award, protests, and modifications add another 12–18 months.
5. By first delivery (3–5 years post-requirement), the threat environment has changed.

This process made sense for hardware platforms with 30-year service lives. It is dysfunctional for software-defined systems, AI-enabled command tools, and autonomous drones where technology generations turn over every 18–24 months.

Other Transaction Authority (OTA): The New Guard's Path In

The mechanism that has opened the door for companies like Anduril, Palantir, and Shield AI is the Other Transaction Authority (OTA), codified at 10 U.S.C. § 4021–4022. OTA contracts:

• Are exempt from the FAR — no sealed bidding, no detailed certified cost or pricing data, no mandatory clause incorporation.
• Can be awarded to non-traditional defense contractors (companies for whom defense is not their primary business and who have not produced a major system for the DoD in the past year).
• Allow prototype agreements (OTA prototype) that can transition directly to production contracts without a new competitive solicitation.
• Move substantially faster: 3–12 months from concept to award is typical.

OTA obligations grew from $4.4B (FY2017) to $22.5B (FY2023) before moderating to ~$18B in FY2025 as the initial OTA wave matured into production contracts. The growth trajectory reflects DoD's explicit acknowledgment that FAR-based acquisition is incompatible with software-era defense technology.

Anduril Industries, for example, has won multiple OTA contracts for its Lattice AI command platform, Sentry Tower autonomous surveillance system, and Ghost drone. None of these required competing against Lockheed or Raytheon under FAR-based sealed bidding — Anduril competed as a non-traditional contractor in accelerated prototype rounds.

The Primes: Lockheed, Raytheon, Northrop — Still Dominant, Increasingly Vulnerable

The traditional primes retain structural advantages that are not easily disrupted:

• Cleared facilities and personnel: Programs classified SAP (Special Access Program) or above require facilities that have been accredited by the defense intelligence community. Building SCIF-accredited facilities takes years.
• Systems integration: Complex weapons systems (F-35, AEGIS combat system, nuclear command/control) require integration of thousands of subsystems. The primes' decades of system-of-systems integration experience is genuinely hard to replicate.
• Political relationships: Defense contracts are embedded in political economies — congressional delegations protect in-district prime contractor facilities regardless of program performance.
• Long-cycle revenue: A contract for F-35 sustainment (Lockheed) or Trident II missile maintenance (Northrop) runs 20–30 years. This generates stable FCF that compounds.

Where the primes are vulnerable:

• Software modernization: All four primes have large legacy software programs (C2 systems, mission planning tools, logistics software) that were built on 1990s–2000s architectures. DoD is increasingly unwilling to fund the primes to maintain these systems and is instead contracting with companies like Palantir (via its Government Platform) to replace them.
• Autonomous systems: None of the primes have a credible internally-developed drone swarm capability competitive with Anduril's Ghost/Fury systems or Shield AI's Hivemind platform. They are responding through acquisition (Northrop acquired OceanMind, L3Harris acquired NOVA Systems) but are behind the technology curve.
• Speed: The primes' overhead structures — compliance, program management bureaucracy, earned value management systems — make them structurally slower than software-first competitors.

Software-First Defense Companies: Palantir, Anduril, Shield AI

Palantir Technologies (PLTR, $50B+ market cap, 2026): Palantir's defense revenue (~$800M+ annually from U.S. government in FY2025) comes primarily from its Gotham platform (intelligence and counterterrorism) and Maven Smart System (Army AI targeting). Palantir is the only software-first defense company with 20+ years of classified program experience — it has the cleared facilities, the trusted workforce, and the program history that qualifies it for sensitive programs the newer entrants cannot access. Its AI Platform (AIP) launched in 2023 is being deployed for both government and commercial users.

Anduril Industries (private, founded 2017, estimated $14B+ valuation post-2024 raise): Palmer Luckey's company has built Lattice (AI command and control software), Sentry Tower (autonomous border/base surveillance), Ghost (VTOL surveillance drone), Fury (autonomous combat air vehicle), and Roadrunner (autonomous counter-drone interceptor). Anduril is structurally positioned to be the Lockheed of autonomous systems if the DoD's shift toward attritable (low-cost, expendable) platforms continues. Its 2024 award for OBSS (Optionally Manned Fighting Vehicle subsystem) and JADC2 contracts signals mainstreaming.

Shield AI (private, ~$2.8B valuation, 2023): Focused on AI piloting (Hivemind platform) for autonomous aircraft. Shield's V-BAT autonomous aircraft is deployed operationally by the U.S. military and multiple NATO allies. Shield's thesis is that AI pilots outperform human pilots in contested electronic environments — the Pentagon's ACE program (Air Combat Evolution) has validated this in training exercises.

The Government Budget: Where the Money Actually Goes

FY2026 DoD budget (enacted at ~$895B):

The technology opportunity is concentrated in RDT&E ($145B) — where OTA mechanisms and prototype contracts are most prevalent — and in the O&M software and IT modernization programs within the $330B O&M budget. The perception that "defense tech" is only about shiny new weapons systems is wrong; the largest software opportunity is modernizing the DoD's enormous legacy IT infrastructure.

Autonomous Systems: Drones, Counter-Drone, and the Ukraine Effect

The Russia-Ukraine war (ongoing) has been the most significant real-world laboratory for autonomous systems doctrine since World War II. Key lessons being incorporated into U.S. procurement:

• Attritable drones: Ukraine's use of Bayraktar TB-2, Shahed lookalike drones, and FPV kamikaze drones demonstrated that mass-produced, low-cost autonomous systems can achieve strategic effects. The U.S. Replicator Initiative ($500M FY2024–2025) aims to field thousands of small autonomous systems.
• Counter-drone: Every base in the Middle East and Europe now requires a dedicated counter-drone layer. Anduril's Roadrunner, Dedrone (acquired by Axon), and L3Harris's VAMPIRE system are all receiving accelerated procurement.
• Electronic warfare integration: Drones without GPS-resilient navigation and EW countermeasures are vulnerable to adversary jamming. Shield AI's Hivemind explicitly addresses GPS-denied navigation.

AI in Defense: Command & Control, Logistics, Intel

The DoD's AI investment is concentrated in three areas:

1. Command and Control (C2): Palantir's Maven Smart System for the Army automates intelligence fusion and targeting recommendation. JADC2 (Joint All-Domain Command and Control) is the $3B+ program to create a unified C2 layer across all services.
2. Predictive Maintenance and Logistics: AI applications predicting component failure before it occurs (reducing costly aircraft groundings) and optimizing supply chain pre-positioning. Palantir's Foundry and SAP S/4HANA are competing for these contracts.
3. Intelligence Analysis: NGA (National Geospatial-Intelligence Agency) and NRO (National Reconnaissance Office) are the largest consumers of computer vision and imagery analysis AI. Palantir, Scale AI, and Maxar are key vendors.

The Investment Case: Public vs. Private Exposure

Public exposure:

• Palantir (PLTR): The only pure-play software-first defense tech company at scale with public liquidity. Trades at ~35–40x 2026E revenue — a premium that reflects its position as the primary proxy for defense AI investment. Earnings inflecting positive.
• Traditional primes at value: Lockheed (LMT), RTX, and Northrop (NOC) trade at 15–20x earnings with strong FCF yields and dividends. They are value plays on defense budget growth, not growth plays on technology disruption.
• L3Harris (LHX): Best-positioned prime for EW, ISR, and communications modernization — overlaps most with where software-first companies compete.

Private exposure (for institutional investors):

• Anduril (latest round at $14B), Shield AI ($2.8B), Joby Defense, Rebellion Defense (acquired by Shield AI 2023), Epirus (directed energy).

Risks: Political, Budget, Procurement Timelines

• Budget risk: Continuing resolution cycles can delay program starts by 6–12 months. A hard spending cap from a budget deal could compress RDT&E and procurement.
• Political risk: Changes in administration affect procurement priorities. The current administration's emphasis on AI and autonomous systems favors the new guard; a reversal could slow OTA growth.
• Procurement timing risk: OTA contracts don't guarantee production. A prototype that wins an OTA award can fail to transition to production if the program gets restructured — a consistent risk for smaller companies dependent on a few programs.
• Clearance and workforce: Security clearance backlogs (currently 600,000+ pending investigations) constrain how fast software companies can hire cleared engineers and access classified program data.

Takeaways for Investors

• The structural shift to software-first defense is real and accelerating: OTA growth, the Replicator Initiative, JADC2, and combat-validated lessons from Ukraine all point in the same direction.
• The primes are not dying, they are compressing: L/M, RTX, and Northrop will sustain revenue from legacy programs for 20+ years. The question is whether they can build or buy the capability to win the next generation of software-defined programs.
• Palantir is the only public equity proxy for defense AI at scale: Its valuation reflects this scarcity premium. Short-term traders will debate the multiple; long-term investors should focus on ARR growth trajectory and classified program wins.
• The private market is where the asymmetric upside is concentrated: Anduril at $14B is building toward a potential public market valuation of $50–100B if its autonomous systems programs achieve planned scale. The risk is procurement timeline slippage.