Why in the News
The core of the United States military industrial complex (MIC), the tie between the armed forces and the firms that supply them, is shifting from traditional hardware makers to software and Artificial Intelligence (AI) firms such as Palantir, Anduril and SpaceX. The shift moves defence power from factories to code, creating new dependencies and a verification gap with fallout for defence industrial bases worldwide.
What is changing in the US military industrial complex?
- Old model: Lockheed Martin, RTX and Northrop Grumman, the Big Three, built few but exquisite multibillion dollar platforms on the logic that superior technology deters rivals.
- New players: Software firms such as Palantir, SpaceX and Anduril, called neo primes, sell code and AI instead of platforms.
- Market verdict: The neo primes’ combined enterprise value (the market value of the whole business) is three times that of the Big Three, and their revenues are nearly eight times smaller.
- Old players persist: Traditional contractors still receive most budget dollars and spend far more on lobbying.
- The takeaway: Investors are betting on software over hardware, even though budgets have not yet followed.
What is driving the shift?
- Inversion of war economics: US and allied defences spent expensive interceptors against cheap drones, an unsustainable cost exchange. The Pentagon responded:
- the Defence Autonomous Warfare Group (DAWG) absorbed the stalled Replicator Initiative, a Biden era plan to field thousands of low cost uncrewed systems quickly;
- DAWG’s projected budget jumps from $225.9 million in FY26 to $54.6 billion in FY27.
- Procurement and capital logic: The old industry used cost plus contracts (the government repays costs plus a margin). Neo primes take fixed price contracts, absorb the upfront costs and earn higher margins by delivering on time.
- Software defined warfare: Code, not platforms, now decides outcomes. A warship’s systems can improve overnight through one download, and a drone can switch from reconnaissance to electronic attack by loading different software.
How has the Pentagon changed the way it buys?
- Anduril contract: The Army awarded Anduril a $20 billion, 10 year enterprise contract in March 2026 that merged more than 120 procurement actions into one.
- Contract consolidation: 14 such contracts replaced 118 individual ones, cutting the count by almost nine in ten. They were signed in eight months, where awards once took seven years.
- Maven Smart System: Palantir’s system became an official programme with locked in funding, the first time a Silicon Valley company is embedded in the defence budget.
What does the new military AI stack look like?
- White House Accord: The Accord on Super Intelligence of 29 September 2026 is a voluntary framework that symbolically brought AI firms into national security.
- May 2026 agreements: The Department of War announced agreements with eight AI companies to deploy on classified networks at impact levels (security grades for handling classified data) 6 and 7.
- Layers of the technology stack: Each layer has named suppliers:
- brain: OpenAI for multilingual intelligence analysis, Google for satellite imagery target recognition and Reflection AI for autonomous combat agents;
- compute: Nvidia’s military grade GPU clusters;
- cloud: Microsoft, Amazon Web Services and Oracle;
- communications: SpaceX’s Starlink and Starshield;
- edge: Palantir and Anduril for edge AI, drone swarms and counter drone systems.
Challenges
- Single points of failure: Palantir’s battlefield systems and SpaceX’s launch capability are new points where one failure disrupts operations.
- Verification gap: The Pentagon’s test and evaluation system, designed around static hardware, lags commercial AI, so unvalidated software defined weapons can fail after a cyberattack.
- Autonomy and accountability: When autonomous agents pick targets, responsibility for a wrong strike is unclear.
Way Forward
- Continuous testing: Redesign test and evaluation for frequently updated software, with validation and cyberattack drills before each deployment.
- Vendor redundancy: Keep alternative suppliers for critical layers such as launch.
- Human control: Keep human decision authority over lethal use of force.
Conclusion
Victory in future wars turns on algorithms, data and code, and commercial AI moves faster than the Pentagon can verify it. The point to watch is whether test and evaluation reform keeps pace with funding for autonomous systems.
About Drone Warfare
- What it is: Use of Unmanned Aerial Vehicles (UAVs), or drones, in combat and surveillance, for precision strikes or intelligence gathering.
- State use: Between 2018 and 2023, states using drones rose from 16 to 40.
- Non state use: Over the same years, non state groups that carried out at least one drone attack rose from six to 91.
- Trends: AI enabled swarms coordinate dozens or hundreds of drones. Eg. Pakistan’s drone swarms during Operation Sindoor. Jammers, lasers and radar detection are the main counters.
Matching Previous Year Question
“[2026] Which of the following statements with regard to drone swarms is/are correct? 1. They use Terahertz band of frequency to communicate with the command centre. 2. Individual drones in the swarm can communicate with other drones in the swarm. 3. GPS Spoofing is a commonly used technique to counter drone swarm attack. (a) 1 only (b) 2 and 3 only (c) 1 and 2 only (d) 1, 2 and 3 ANSWER: B”
