The Valley of Death Is a Structural Problem
The Department of Defense spends billions each year funding early-stage technology development through programs like SBIR, STTR, AFWERX, and the Defense Innovation Unit. The stated goal is straightforward: identify promising commercial and dual-use technologies, prototype them for military applications, and transition them into production contracts that deliver capability to the warfighter.
The reality is that most of these technologies never make it to production. Air Force SBIR Phase II-to-Phase III transition rates sit at roughly 8.8%. For every eleven companies that successfully demonstrate a working prototype under a Phase II contract, fewer than one will secure the follow-on production funding needed to deliver their technology at scale. The rest enter what the defense innovation community calls the “valley of death,” a gap between prototype completion and production contracts where promising companies burn through capital, lose technical talent, and eventually pivot away from defense entirely.
This is not a new problem, but it is getting worse. The defense industrial base has shrunk by 27.6% over the past decade, and new entrants seeking to work with DoD have dropped roughly 80% over the past 15 years. The Pentagon’s own State of Competition report documented the consolidation from 51 prime contractors in 1993 to just five today, with tactical missile suppliers shrinking from 13 to 3 and fixed-wing aircraft manufacturers from 8 to 3. The industrial base that the innovation ecosystem is supposed to feed is contracting faster than the ecosystem can replenish it.
Where the Innovation Dollars Go
The federal government has built an increasingly large infrastructure to fund defense innovation. The problem is not the size of the investment. It is what happens after the check clears.
DoD is the largest federal SBIR/STTR participant, awarding approximately $1.8 billion per year through 14 separate components, more than all other federal SBIR agencies combined. Government-wide SBIR/STTR spending rose 31% from $4.7 billion in FY2020 to $6.3 billion in FY2023, with DoD and defense agencies accounting for $5.6 billion, or 91%, of total spending.
AFWERX, the Air Force’s innovation arm, has been particularly aggressive. Since 2019, the organization has awarded over 10,400 contracts worth more than $7.24 billion to strengthen the defense industrial base. In FY2024 alone, 470 companies transitioned to Phase III contracts totaling $1.44 billion. AFWERX operates with an annual budget of approximately $1.4 billion.
The Defense Innovation Unit, established in 2015 to connect Silicon Valley technology with military requirements, has made 450 prototype awards from FY2016 through FY2023. Of completed prototypes, 51% transitioned to production, a rate significantly higher than the broader SBIR ecosystem but still meaning that nearly half of all DIU prototypes never reached the warfighter. The 62 prototypes that did transition carried a combined contract ceiling value of $5.5 billion.
Other Transaction Authority agreements, the primary contracting mechanism for these innovation programs, have grown dramatically. OTA obligations expanded from $1.8 billion in FY2016 to over $18 billion in FY2024, roughly a tenfold increase in eight years. By FY2023, DoD had more than 1,200 active OTAs worth more than $23.6 billion.
The money is flowing. The transition is not.
Why Companies Get Stuck
The valley of death is not primarily a funding problem. It is a compliance, credentialing, and institutional problem. Companies that successfully build a working prototype under an OTA or SBIR contract face an entirely different set of challenges when they attempt to move into the traditional acquisition system.
Cybersecurity compliance. CMMC Level 2 certification, required for handling Controlled Unclassified Information, costs small businesses (under 50 employees) between $75,000 and $130,000 in first-year compliance costs, with 20 to 30% annual maintenance costs on top. Assessment fees alone run $30,000 to $70,000. Given that over 70% of the defense supply chain consists of small and mid-size businesses, CMMC compliance represents a structural barrier that disproportionately affects exactly the companies the innovation ecosystem is trying to cultivate.
Export controls. ITAR compliance adds another layer of cost and complexity. First-year compliance for a startup runs approximately $35,000, covering DDTC registration, gap assessments, policy development, training, and technical controls. Ongoing costs average roughly $12,000 per year. The penalties for getting it wrong are severe: up to $1 million per violation, 20 years imprisonment, and permanent debarment. For a commercial technology company accustomed to selling software globally without export restrictions, ITAR compliance requires a fundamental rethinking of how the company operates.
Security clearances. Many defense programs require cleared personnel, but the clearance process itself creates a chicken-and-egg problem. Average end-to-end processing time for a security clearance is 243 days as of Q3 FY2025. The backlog reached approximately 300,000 cases by the end of FY2024, the highest level since Q4 FY2019. Across the country, 500,000 to 700,000 positions requiring security clearances remain unfilled. A startup that wins a contract requiring cleared engineers cannot begin work until the clearances process, which may take eight months or longer.
Past performance requirements. Perhaps the most pernicious barrier is the Catch-22 of past performance evaluation. Federal solicitations routinely require bidders to demonstrate relevant past performance on similar contracts. Companies emerging from SBIR Phase II or DIU prototype programs have exactly zero relevant production contract past performance, which makes them noncompetitive against incumbents regardless of how superior their technology may be. The FY2026 NDAA addressed this partially through Section 824, requiring DoD to issue guidance on accepting past performance references from commercial contracts and using alternative evaluation methods when a requirement has no precedent. Whether that guidance produces meaningful change remains to be seen.
Acquisition timelines. Major defense acquisition programs now take almost 12 years to deliver initial capability, and that timeline has increased by 30%, resulting in average delays exceeding two years. A dual-use technology company cannot survive a 12-year sales cycle. Most venture-backed startups measure their runway in months, not decades.
The Companies That Made It Across
Despite the structural barriers, a small number of companies have successfully crossed the valley of death and built substantial defense businesses. Their paths offer lessons, though not necessarily replicable ones.
Palantir Technologies spent more than a decade and hundreds of millions of dollars fighting its way into the defense market. The company’s breakthrough came not through the traditional acquisition system but through a series of legal challenges and direct engagement with operational military units. Today, Palantir’s U.S. government revenue reached $486 million in Q3 2025 alone, representing 52% year-over-year growth. The company holds a $10 billion Army enterprise data agreement and its Maven contract expanded to approximately $795 million in 2025. Palantir succeeded, but it took a publicly traded company with billions in commercial revenue and the resources to litigate procurement disputes to do it.
Anduril Industries took a different approach, building hardware and software systems purpose-designed for defense applications while deliberately structuring its business to navigate the acquisition system from day one. The results have been extraordinary: $2.1 billion in revenue in 2025, up 110% from $1 billion in 2024, which itself was up 138% from $420 million in 2023. Anduril’s valuation reached $61 billion in its May 2026 Series H. The company secured a $20 billion ten-year Army enterprise contract in March 2026, consolidating over 120 procurement actions into a single vehicle. Anduril’s advantage was founder experience (Palmer Luckey had both capital and defense community credibility), substantial venture backing, and a deliberate decision to build compliance infrastructure before pursuing production contracts.
Shield AI demonstrates a third path, building autonomous drone capability and scaling through a combination of military contracts and operational deployments. The company reached a $12.7 billion valuation in its March 2026 Series G, up from $5.6 billion in 2025 and $2.8 billion in 2023. Revenue grew roughly 64% year-over-year to approximately $267 million in 2024, with projections exceeding $540 million for 2026. Shield AI secured a V-BAT Coast Guard IDIQ worth up to $198.1 million and has competed for up to $800 million in Navy ISR contracts. The company’s V-BAT system has been operationally deployed in Ukraine, completing 35+ missions and identifying over 200 Russian targets.
The common thread across all three is not technology. It is capital, patience, and institutional knowledge. Each company raised billions in venture funding, built dedicated government relations and compliance teams, and had founders or senior leaders with deep defense community relationships. These are not resources available to the typical SBIR Phase II company.
What the System Is Doing About It
The Pentagon has recognized the transition problem and launched several initiatives to address it.
The Replicator Initiative, launched in August 2023, aimed to field thousands of autonomous systems by August 2025. The results were mixed: DoD delivered “hundreds” rather than “thousands” by the deadline, with additional units on contract. Replicator demonstrated that the Pentagon can accelerate procurement when leadership prioritizes speed, but it also highlighted how difficult it is to sustain that urgency across the bureaucracy.
The Chief Digital and Artificial Intelligence Office (CDAO) has taken over AI acquisition oversight across the department, with its FY2025 budget reaching $139.9 million, a dramatic increase from $10.3 million in FY2022 when the office first assumed AI oversight responsibilities.
Legislative reforms have been incremental but meaningful. The FY2026 NDAA included Section 1826, exempting nontraditional defense contractors from certified cost and pricing data requirements, FAR Part 31 cost principles, and certain business system requirements. The simplified acquisition threshold was raised to $350,000 effective October 2025, relaxing past performance and compliance requirements for smaller contracts.
Private capital has responded to the opportunity, even as public acquisition reform lags. Defense tech venture capital hit a record $49.9 billion across 966 deals globally in 2025, representing nearly 8% of all global VC funding. The number of VC firms investing in defense grew from fewer than 100 in 2017 to more than 300 by 2024. NATO-country startups raised approximately $9.1 billion in the first nine months of 2025, up from $6.5 billion in all of 2024.
The capital is flowing because investors see a structural mismatch between defense spending growth and the shrinking industrial base. The question is whether the acquisition system can evolve fast enough to absorb the innovation that capital is funding.
What Dual-Use Companies Should Actually Do
The valley of death is real, but it is not inevitable for every company. The companies that successfully transition from prototype to production share several characteristics that are worth studying.
Start compliance early. The single most common mistake dual-use companies make is treating compliance as a post-award problem. CMMC certification, ITAR registration, facility clearance applications, and CUI handling procedures should begin during Phase II, not after a production contract is awarded. The 243-day average clearance processing time means that waiting until contract award to start the security clearance process guarantees months of delay before work can begin.
Build past performance deliberately. Companies cannot wait for large production contracts to accumulate past performance. Subcontracting under established primes, performing on task orders under IDIQ vehicles, and documenting commercial project performance in CPARS-compatible formats all build the evaluation record that source selection boards require.
Use OTAs as a bridge, not a destination. Other Transaction Authorities are valuable prototyping mechanisms, but they are not a substitute for learning how to operate within the FAR-based acquisition system. Companies that treat OTA contracts as their permanent business model will struggle when programs transition to traditional contract vehicles.
Plan for the timeline. Even with OTA acceleration, the path from prototype to production typically spans 18 to 36 months. Dual-use companies need capital planning that accounts for this gap, which often means raising dedicated defense transition funding or structuring commercial revenue streams that can subsidize the defense business during the transition period.
Find the right primes. Strategic teaming with established defense contractors can provide the compliance infrastructure, past performance, and contract vehicle access that dual-use companies lack. The most effective teaming arrangements are those where the technology company brings genuine capability differentiation and the prime brings institutional access and compliance infrastructure.
The Real Innovation Gap
The defense innovation gap is not a technology problem. The United States has the most dynamic technology sector in the world, and American companies are building capabilities in autonomy, artificial intelligence, advanced manufacturing, cybersecurity, and space systems that are directly relevant to national security.
The gap is institutional. The acquisition system was designed for a world of five prime contractors building large platforms over decades-long timelines. It was not designed for a world where a three-year-old startup with 50 employees can build an autonomous drone system that outperforms legacy platforms at a fraction of the cost. Until the acquisition system catches up to the innovation ecosystem it is supposed to serve, the valley of death will continue to claim companies and capabilities that the warfighter cannot afford to lose.
The $49.9 billion in defense tech venture capital deployed in 2025 represents a bet that the gap will close. The 8.8% SBIR transition rate suggests it has not closed yet.