Defence AI Watch #1: Grok, Iran and the Defence-AI Supercycle — PLTR, KTOS, BBAI and PDYN

The Defence AI Supercycle: Why Now

The strongest available baseline comes from SIPRI: world military expenditure reached roughly $2.887 trillion in 2025, a new record and the eleventh consecutive year of growth. This is not just about larger budgets; it is a structural transformation in how militaries plan, gather intelligence, protect networks, coordinate drones and make real-time decisions. Artificial intelligence is no longer only a future promise. It is increasingly becoming part of the operational infrastructure of defence.

The Middle East conflict has accelerated experimentation and demand for low-cost systems, decision software and tactical autonomy. The new development on June 16, 2026 is that the discussion is no longer limited to Claude, Maven or Palantir: according to reports published today, government documents also cite the use of an xAI Grok model in support of U.S. bombing missions in Iran, with reference to the deployment of more than 2,000 munitions against roughly 2,000 targets over 96 hours. Reuters had already documented the June 9-10 wave of renewed U.S. strikes against Iran; the Grok disclosure adds a different layer because it connects a commercial generative-AI model directly to Pentagon operational capacity in an active theatre.

In this context, four Nasdaq-listed companies are building the architecture of intelligent defence brick by brick: Palantir (PLTR) with data and decision-making, Kratos (KTOS) with physical autonomous systems, BigBear.ai (BBAI) with intelligence analysis, and Palladyne AI (PDYN) with tactical swarm autonomy. They have very different profiles in terms of market cap, development stage, and revenue model. Defence AI Watch is Merlintrader’s monthly column tracking them together, systematically.

PLTR — Palantir Technologies: The Data Engine

The Model

Palantir is the data analytics company founded in 2003 with CIA backing. Today it operates on two pillars: Gotham (government and defence) and AIP (AI Platform, commercial). The model is simple on paper — sell access to a platform that integrates heterogeneous data and builds AI models on top — but extremely difficult to replicate because it requires years of integration with customers’ existing systems.

Q1 2026 Numbers

The quarter reported in May 2026 was exceptional: total revenue of $1.633 billion, up 85% year-over-year and 16% sequentially. The most relevant figure for this column was U.S. Government revenue, up 84% YoY to $687 million. U.S. Commercial grew even faster, up 133% YoY to $595 million. Palantir closed $2.41 billion of total contract value in the quarter and raised FY 2026 revenue guidance to $7.650-$7.662 billion.

The correct metric to watch is not a generic “$4.38 billion total backlog” figure, but U.S. Commercial remaining deal value, which reached $4.92 billion. On profitability, Palantir reported $871 million of GAAP net income in Q1 2026 and $8.0 billion in cash, equivalents and short-term U.S. Treasuries. This changes the risk profile: PLTR remains a high-valuation stock, but it is not a cash-burning story anymore.

The Thesis

PLTR is no longer a bet — it is an almost natural monopoly position in military data analytics. Replacing Gotham in a military organisation that has built years of workflows on top of it is practically impossible. The main risk is valuation: at $134 per share with still-elevated P/S multiples, any growth slowdown can produce a violent correction. But the business direction is crystal clear.

KTOS — Kratos Defense: Physical Systems

The Model

Kratos builds physical autonomous systems: target drones (used to test missile defence systems), rockets, C5ISR systems, and most importantly the low-cost combat drone programme XQ-58A Valkyrie. Kratos’s core idea is to manufacture military drones like consumer products — low unit costs, mass production, expendable. A paradigm shift from traditional weapons systems costing hundreds of millions each.

Q1 2026 Numbers

Revenue was $371 million in Q1. Kratos reported strong order momentum, with its KGS segment producing a 1.8:1 book-to-bill ratio, funded backlog of approximately $1.635 billion and a bid/proposal pipeline of about $14.3 billion. FY 2026 guidance was raised to $1.70-$1.76 billion in revenue, with adjusted EBITDA expected in the $170-$176 million range. The setup remains attractive, but execution matters: defence hardware growth is not only about orders, it is also about manufacturing capacity, delivery timing and margin control.

The Thesis

Kratos is the only listed player that builds both the physical drone and the command systems needed to fly it autonomously. With the Valkyrie programme, Kratos has positioned itself as the reference supplier for the U.S. Air Force “loyal wingman” doctrine — autonomous drones operating alongside piloted fighters. The defence supercycle (rising NATO budgets post-Ukraine, global rearmament, Middle East conflict) generates orders almost automatically. The main risk is execution: more contracts means more operational complexity.

BBAI — BigBear.ai: Intelligence Analysis

The Model

BigBear.ai operates in a precise niche: it provides AI analytics platforms for intelligence, surveillance, and national security. Its main products are computer vision systems, predictive analytics, and decision workflows for government agencies and the intelligence community. It is a much smaller company than the other three — but its positioning in classified contracts makes it potentially difficult to value from the outside, which is both a risk and an opportunity.

Q1 2026 Numbers

Revenue was $34.4 million in Q1 2026, down 1% from $34.8 million in Q1 2025. The positive side came from gross margin and backlog: gross margin expanded to 34.0% from 21.3% a year earlier, backlog increased 14% sequentially to $281.9 million, and the company reported roughly $75 million of quarterly wins, including more than $60 million in national security. That package included a $53 million sole-source classified award with an intelligence customer not publicly identified. FY 2026 guidance remains $135-$165 million.

The Thesis

BBAI is the most speculative bet of the four. The model works — contracts arrive and backlog grows — but profitability is still distant and dependence on a limited number of government customers represents significant concentration risk. The main catalyst for a rerating is a large-scale contract with a high-profile federal agency, or the expansion of Ask Sage (a conversational AI assistant for government use) that could open a broader market. It is therefore a high-beta AI-defence small-cap to monitor, not a business that is already fully de-risked.

PDYN — Palladyne AI: Tactical Autonomy

The Model

Palladyne AI is the smallest and most specialised company in the group. Born as Sarcos Technology (industrial robotics) and repositioned on autonomous AI for defence, it produces three tightly integrated components: SwarmOS (software coordinating UAV swarms in GPS-denied and degraded-communications environments), Gremlin-X (the tactical-autonomy and UAV/munition-integration family), and IntelliSwarm (swarm/autonomy logic integrated into defence-compatible avionics hardware).

The Moment: Ivy Mass and NGC2

In June 2026, Palladyne AI announced that SwarmOS had been deployed during Ivy Mass, a U.S. Army 4th Infantry Division exercise. The release describes autonomous capabilities for modern Army combat operations “at the edge” and integration with the NGC2, or Next-Generation Command and Control, prototype ecosystem. The difference versus a conference demo matters: Ivy Mass is a real military exercise environment, spanning multiple operational domains and putting pressure on coordination among sensors, operators and platforms.

Added to this is the June 8, 2026 partnership with Israel Aerospace Industries to manufacture, integrate and market HARPY, HAROP and Mini HARPY systems in the United States. The agreement gives Palladyne exclusive U.S. production and marketing rights for those loitering-munition systems, while IAI provides engineering support and key subsystems.

Q1 2026 Numbers

Revenue was $3.54 million in Q1, more than double the $1.71 million reported in the same period of 2025. The growth rate is more relevant than the absolute number: this is still an early-stage story. GAAP net loss was roughly $12.6 million, while cash, equivalents and marketable securities totaled $43.7 million as of March 31, 2026. Backlog was approximately $17 million, including roughly $7 million of new quarterly contract awards net of revenue recognized. FY 2026 revenue guidance remains $24-$27 million.

The Thesis

PDYN is the highest-risk/reward story of the group. If SwarmOS becomes the standard software for UAV swarm coordination in the U.S. Army — and Ivy Mass is a step in that direction — the rerating potential is very significant. The current market cap is negligible relative to the TAM (total addressable market) for military autonomous systems. The risk is dilution: with limited cash and still-small revenues, PDYN will almost certainly need new capital over the next 12-18 months. Any dilutive capital raise can compress the price in the short term.

Comparative Snapshot — June 16, 2026

Intraday market data captured on June 16, 2026 around 17:47 UTC. Prices and percentage moves can change during the session.

Possible 12-Month Scenarios

The four stocks are at very different stages of development. Below are the main scenarios for each, without operational recommendations.

PLTR

KTOS

BBAI

PDYN

Cross-Sector Risks

Merlintrader Bottom Line

Defence AI is not one single theme: it is a value chain. The Grok/Iran disclosure makes that chain more visible: foundation models, data-fusion platforms, targeting workflows, sensors, drones, munitions and autonomous systems are beginning to integrate into a single operational architecture. PLTR controls the data/decision layer with scale, margins and cash; KTOS controls a physical layer of autonomous systems and attritable drones; BBAI remains an intelligence-analytics small-cap with a stronger balance sheet but growth still to prove; PDYN is the earliest and most asymmetric profile, tied to SwarmOS validation, loitering munitions and the ability to convert military demonstrations into recurring orders.

The most important point for readers is not to pick a theoretical “winner.” It is to separate business quality, revenue maturity, balance-sheet strength, valuation and catalysts. At this stage, PLTR and KTOS are the more institutionalized stories; BBAI and PDYN are more speculative and much more dependent on execution, government awards and access to capital. Defence AI Watch will track that evolution month after month, separating verified facts, market narrative and real risks.