A recent paper from Stanford's Gordian Knot Center, Defense Tech's Gold Rush by Brian Katz, lays out the structural challenge facing the sector with unusual clarity. The core argument: defense tech companies face not one but two distinct "valleys of death," and the second one is where most of today's valuations will be tested.

Valley One Is Familiar Territory

The leap from prototype to production — navigating OTAs, securing an institutional sponsor, getting through ATO accreditation. Recent reforms have narrowed this gap. DIU's transition planning requirements, commercial-first acquisition mandates, and the emerging Portfolio Acquisition Executive (PAE) construct are all moving in the right direction.

Valley Two Is Where Things Get Harder

This is the gap between production and programming, between having a contract and being written into the Program Objective Memorandum (POM), the five-year budget document that determines which capabilities the Department actually funds and sustains long-term.

The distinction matters because production contracts, even large ones, are often just permission to compete again next year. An IDIQ with an impressive ceiling authorizes buying, but it doesn't obligate it. A company can be operationally deployed, praised in briefings, and renewed annually while remaining entirely outside the enduring force structure. Katz calls this "production purgatory," and the term fits.

Why Valley Two Is an Investor Problem

The paper makes a point that defense investors should sit with: current valuations across the sector rest on a critical assumption, that operational relevance will translate, with reasonable speed, into institutional adoption. PEO ownership, POM inclusion, multi-year funding. For a significant share of companies, that conversion will take longer than standard venture timelines allow, if it happens at all.

Programming decisions are zero-sum. For a new capability to get funded, something else has to lose funding. That means venture-backed companies aren't just competing on technical merit; they're competing against sunk costs, existing training pipelines, sustainment models, and Service identities tied to legacy platforms. The paper frames this as startups being "rifle shots" versus primes operating as "mutual funds" — diversified, able to absorb risk across missions and decades. The POM isn't designed to reward the best-performing system. It allocates future-year risk. Financial durability becomes a proxy for institutional safety.

The practical implication: the difference between a 3x outcome and a 30x outcome in defense rests less on product quality than on institutional durability.

What This Means for Defense Capital Allocation

Katz outlines several recommendations for investors that align with how we think about deals at Novatus:

Production-to-programming should be modeled as a multi-year process, not a discrete milestone. Companies reaching early production aren't at the finish line; they're entering the most institutionally complex phase of their lifecycle. Traction without a named Service sponsor or plausible POM pathway is fragile and should be valued accordingly.

Institutional fluency matters as much as technical capability. Understanding PPBE cycles, portfolio politics, and budget mechanics isn't a nice-to-have — it's a core asset. The teams that understand how the Department actually programs and budgets are the ones positioned to survive Valley Two.

Acquisition by a prime isn't failure. The paper is refreshingly direct on this point: for many companies, integration into a prime's portfolio is the fastest and most realistic route to scale. Designing for that outcome early is strategy, not surrender.

The Structural Case Remains Intact

None of this undermines the fundamental demand signal. Ukraine demonstrated the decisive role of AI, autonomy, and attritable systems. China's doctrine of intelligentized warfare targets exactly the kind of static, legacy architectures the Department is trying to move away from. The alignment between battlefield evidence, policy intent, and private capital is real and structural — not cyclical.

But structural demand doesn't automatically convert into durable company-level outcomes. The institutional machinery between "this works" and "this is part of the force" remains the binding constraint. The 2026 National Defense Strategy laid out where the Department intends to spend — this paper explains why getting from intent to institutionalized spending is harder than most investors appreciate.

The full paper is worth reading for anyone deploying capital into the sector — it's one of the clearer-eyed assessments of where the real risk sits.