Pentagon Doubles Down on Exquisite Weapons|Kratos Upgraded on Mass Production Bet

The Budget That Contradicts Itself

The largest proposed Pentagon budget increase since the Korean War landed this week, and the immediate instinct was to buy defense stocks across the board. That instinct is not wrong. But it is incomplete.

The proposed FY2027 budget calls for a 44% spending increase, with weapons procurement set to more than double over two years. JPMorgan called it a transformation of the defense industrial base. On the surface, a rising tide lifts all ships.

Here is where the tension begins. The same week this budget dropped, Jefferies upgraded Kratos Defense to Buy, citing a 14 billion dollar opportunity in hypersonics and rocket motors. Kratos is not a battleship company. It builds affordable, attritable unmanned systems — the kind designed to be expendable in volume. And yet the Pentagon budget it is supposed to benefit also funds a new battleship class, space-based missile interceptors, an F-47 fighter potentially costing 300 million dollars per unit, and a B-21 bomber at over 600 million each.

Two very different theories of modern warfare just got funded simultaneously. That contradiction is the story.

Cheap Drones vs. Expensive Interceptors

The Iran conflict exposed something that defense planners had theorized but now have empirical data on. Iran's Shahed drones cost between ten thousand and fifty thousand dollars each. The US interceptors used to shoot them down cost millions per shot. That exchange ratio is mathematically unsustainable at scale.

Before that conflict, the US was procuring roughly 90 Tomahawk cruise missiles per year. The maximum production rate was 2,330 per year. The gap between what was being bought and what the industrial base could theoretically produce was not a secret — it was a budget choice. A choice that, in a long-war scenario, becomes a strategic liability.

General David Petraeus, former CIA director, framed this plainly. He called Ukraine the most important military industrial complex in the free world. Not because of its size, but because of its tempo. Ukraine pushes software updates in under a week. Hardware changes happen every few weeks. It is selling drones to Gulf states to defend against the very Iranian systems that exposed the US intercept-cost problem.

Petraeus's argument is that the cheap mass-production model is what the future requires — not the exquisite weapons model. And yet the 44% budget increase leans heavily into exquisite. That is not an oversight. Pentagon officials acknowledged the lesson from Ukraine: field both. But when procurement dollars more than double and the headline items are battleships and stealth bombers, the balance of that "both" matters enormously for allocating exposure within defense.

CrowdStrike co-founder Dmitri Alperovitch added another layer. NATO's production capacity and supply chains are not built for a long war. Russia, despite a significantly smaller GDP, outproduces NATO in artillery, armored vehicles, glide bombs, and drones. The exquisite-vs-mass debate is not academic. It maps directly to which companies in the defense supply chain are positioned for the next five years versus the next five months.

The Reversal Most Are Missing

Here is the point that gets lost in the noise around a Jefferies upgrade or a budget headline.

Kratos at 88 times operating cash flow — more than double its five-year average — is being valued as if the attritable, high-volume future is already locked in. The upgrade is well-reasoned. The 14 billion dollar hypersonics and rocket motors opportunity is structural. But the valuation already prices in a scenario where the budget priorities align with Kratos's model. If the exquisite-weapons allocation dominates actual spending, the volume that justifies 88x never materializes.

The four defense executives Fortune interviewed anonymously raised a related concern that the market has largely set aside. The US carries 39 trillion dollars in national debt. Monthly interest payments are running at 88 billion dollars — roughly equal to the combined defense and education budgets. A 44% Pentagon spending surge does not exist in a fiscal vacuum. Sustainability questions do not kill a bull cycle. But they set a ceiling on how long the cycle runs and how much political capital survives a recession or a bond market event.

This is the reversal card: the budget surge is real, and the defense cycle is genuine. But the companies best positioned for the next war — mass production, software-defined, attritable — are not the same as the companies best positioned for how the current budget is actually weighted. Knowing which basket is which changes the analysis materially.

Palantir's Battlefield Credential vs. the Burry Problem

Palantir's situation runs on a separate track, but it converges at the same question: what does the Pentagon actually buy when it buys AI?

Palantir's AI Maven system was reportedly used in US military operations during the Iran strikes. That is not a product pitch — it is a live deployment record. Ten consecutive quarters of accelerating revenue growth, with the US commercial segment expanding 137% year over year, suggests the platform is compounding, not plateauing.

Michael Burry's argument is different. Anthropic's annualized revenue surged from 9 billion to 30 billion dollars in a matter of months. Burry read that as displacement pressure on Palantir's commercial AI position. The market responded — Palantir dropped 7.3% on volume running 82% above its three-month average. The stock is now down 26% from where Burry first made this call in early November.

The counterargument is structural rather than defensive. Palantir does not sell a large language model. It sells an operating system for decision-making inside classified and commercial environments where data sovereignty, auditability, and integration with existing infrastructure are non-negotiable. The Pentagon has already flagged restrictions on certain Anthropic technology for defense use. That is not a minor footnote — it is a procurement constraint that Anthropic has to solve before it competes for the contracts Palantir already holds.

The more pointed risk for Palantir is not Anthropic. It is whether the pace of AI-enabled warfare development from China forces the Pentagon to diversify its AI vendor base faster than any single platform can adapt. Defense executives have said explicitly that China's trajectory on AI in warfare is a primary concern. If that pressure accelerates, the question is whether Palantir's moat is deep enough to hold exclusivity — or whether it becomes one of several platforms in a fragmented ecosystem.

Evidence leans toward the moat holding in the near term, but only if Palantir continues converting its battlefield deployment record into locked contracts before the competitive field widens. The 26% drawdown has compressed the valuation relative to its growth rate, which changes the risk profile from what it was in November.

Scenario Branching and the Single Axis

The convergence point across all of these threads is the same: the 44% budget surge is not a uniform tailwind. It is a directional bet on a specific theory of warfare, and that theory is internally contested.

If the exquisite-weapons allocation dominates actual procurement — battleships, F-47s, space interceptors — the companies most exposed to high-unit-cost, low-volume programs benefit. The companies building volume-production platforms, autonomous systems, and software-defined warfare tools benefit less than the headline numbers imply. Kratos's upgrade thesis weakens. Palantir's defense moat matters more in a software-forward environment — but budget dollars flowing to hardware reduce the urgency.

The upside path for both names runs through a different scenario: Congress and the Pentagon absorb the Ukraine and Iran lessons more fully in actual procurement allocation, not just in stated doctrine. In that case, the 14 billion dollar hypersonics and rocket motors opportunity Jefferies identified for Kratos is not speculative — it is the logical recipient of shifted dollars. Palantir's AI Maven platform scales from operational use into formal program-of-record status, locking in multi-year contract structures that Anthropic's current clearance profile cannot easily displace.

The downside path is fiscal constraint meeting strategic incoherence. A bond market that reprices US debt sustainability, or a Congress that cannot sustain 44% budget growth across multiple cycles, does not kill the defense sector. But it forces prioritization. And a forced prioritization between exquisite and mass-production is a decision that has not yet been made clearly — and that ambiguity is the single largest unpriced risk in the current defense bull thesis.

The defense cycle is real. The budget commitment is the largest in over seventy years. But the allocation within that cycle — not the aggregate size — is what determines which companies in the stack actually compound from here.