The United States continues to build the best weapons. It is less effective at producing fast, adaptable, and affordable systems at scale. Nowhere is this clearer than with drones: Ukraine has demonstrated the battlefield gains from rapidly developing and deploying low-cost systems, while US procurement still favors expensive, slow-moving platforms optimized for a small number of high-end scenarios.
Current US procurement rules reward caution, complexity, and incumbency. Cost-plus contracts insulate prime contractors from downside risk. Detailed ex ante specifications lock the military into “gold plated” designs that take decades to deliver. New entrants face formidable barriers—from classification to compliance to the risk that the government will change course after firms have sunk large fixed costs.
The result is a persistent pattern of delay, cost escalation, and missed opportunities. The F-35 fighter aircraft program is the most visible example, but the same logic increasingly applies to emerging technologies such as drones and autonomous systems, where speed and iteration matter most.
We argue that the United States government should adapt a tool it used successfully during the COVID-19 crisis: advance market commitments. Properly designed, advance market commitments can realign incentives in defense procurement, encourage entry and experimentation by new firms, and shift the system away from perfectionism toward rapid, deployable capability.
How advance market commitments work
Advance market commitments (AMCs) reduce uncertainty for potential entrants by having the government commit to buying a specified quantity at a set price if a product meets clearly defined minimum performance thresholds. This approach was central to Operation Warp Speed and the development of the COVID vaccine.
Before he became Stanford’s president, Jonathan Levin co-authored a piece with a Nobel laureate, Michael Kremer, that argued that advance market commitments solve a fundamental incentive problem faced by firms:
Firms invest in R&D and capacity before bargaining with purchasers over price and quantity. Purchasers expropriate some investment returns in bargaining, leading the firm to underinvest. The firm may not develop the vaccine at all.
By guaranteeing demand for COVID vaccines that met basic efficacy and safety standards, the US government encouraged firms to invest in parallel development and to scale production before the vaccines’ efficacy was fully known.
Defense procurement shares many of the same structural challenges as vaccine development: high upfront costs, uncertain research and development outcomes, and the risk that firms will be exposed if government priorities change after investments are made. Advance market commitments would trigger purchases once performance is good enough to justify deployment, while preserving the ability to learn and adjust over time.
Recent policy developments reflect a growing recognition of these challenges. Defense acquisition guidance has increasingly emphasized speed, experimentation, and engagement with nontraditional firms, and recent National Defense Authorization Acts have expanded authorities intended to accelerate prototyping and fielding, particularly for autonomous and so-called attritable systems.
These reforms largely focus on process flexibility rather than demand-side commitment. Advance market commitments complement these efforts by providing credible purchase guarantees that make it profitable for new entrants to invest, scale, and experiment in the first place.
Drones: an illustration
Consider how an advance market commitment could be applied to drone procurement.
Step 1: A credible commitment. The Pentagon announces that it will purchase, for example, ten thousand drones at a fixed unit price if minimum performance criteria are met. The commitment is backed by appropriated funds and includes options to scale up orders for superior performers.
Step 2: Competitive entry with screening. Participation is open, but not indiscriminate. Firms must demonstrate technical competence, submit prototypes, and meet security and sourcing requirements. This ensures serious competition without opening the process to unqualified entrants.
Step 3: Minimum viability thresholds. Rather than exhaustive specifications, the government sets “good enough” benchmarks: reliability, cost ceilings, payload, range, and basic interoperability. These thresholds are intentionally modest. The goal is deployability, not perfection.
Step 4: Tiered rewards, not gold-plating. Performance above the minimum is rewarded through higher prices, larger orders, or co-investment in scaling. This creates incentives for improvement without forcing all bidders into over-engineered designs that delay fielding and raise costs.
Step 5: Safeguards. Contracts would include sunset clauses, independent audits, and penalties for noncompliance with sourcing or security rules. Firms bear research and development risk but can be confident that if they meet the agreed-upon performance targets, they will be paid. Threshold manipulation can be mitigated through independent oversight and transparent metrics.
Rewards for experimentation
This incentive system encourages experimentation and entry by firms that would never survive a traditional procurement cycle. The Pentagon could propose a pilot program to test whether these incentives foster significant innovation and military entrepreneurship for a class of weapons such as drones.
Recent guidance from the Food and Drug Administration on the use of Bayesian statistical methods offers a useful analogy for how the Pentagon could structure learning in an advance market commitment pilot. The FDA has emphasized that regulators need not wait for absolute certainty before acting in high-stakes environments. Instead, decisions can be triggered once evidence crosses a confidence threshold that justifies deployment, while remaining open to continued learning as new data arrive.
A key feature of this approach is the use of skeptical priors. In settings where many similar products have failed, or where dramatic performance improvements are unlikely, regulators appropriately begin with caution. New evidence must earn its way into belief updating. This logic maps naturally onto defense procurement, where many startup firms propose systems that resemble prior platforms that underperformed or failed to scale. Beginning with skepticism is not a flaw; it is a disciplined response to historical experience.
In a drone procurement pilot using advance market commitments, the Pentagon would publicly commit to this structure ex ante. Firms would be informed that evaluations begin from a skeptical baseline shaped by prior systems and recent field experience. They would also be told which performance dimensions matter most and what minimum evidence thresholds must be crossed to trigger procurement at different funding tiers. As testing proceeds, the government would update its assessment as new data arrive, increasing confidence as performance improves across relevant margins.
Crucially, this process would not require disclosing precise internal probability estimates. Instead, firms would receive structured, directional feedback about how close they are to meeting procurement thresholds and which additional data would most influence approval. For example, a firm might be informed that endurance and reliability testing has materially increased confidence, while survivability or cost remains the binding constraint for scale-up. This mirrors regulatory practice in other high-stakes domains, where transparency about evaluation criteria and belief updating encourages investment without inviting gaming.
Embedding this Bayesian learning framework within an advance market commitment pilot would formalize what is often implicit and ad hoc in defense acquisition. It would reassure entrants that evidence will be evaluated consistently, that success will be rewarded, and that skepticism is not immovable. At the same time, it would allow the Pentagon to move faster by acting when confidence is sufficient rather than waiting for perfection. In environments where adversaries iterate rapidly and uncertainty is unavoidable, disciplined belief updating is not a concession—it is a strategic advantage.
Aligning incentives
Recent Hoover Institution research by Dan Berkenstock and Jon Chung has emphasized that procurement outcomes depend not only on formal rules but also on the incentives facing the officials who implement them. Advance market commitments will succeed only if the public officials responsible for administering them are rewarded for disciplined risk-taking rather than punished for experimentation.
One implication is the need for clearly designated internal champions—program executives or acquisition leads—whose careers benefit from successful fielding, rapid scaling, and competitive entry. Rather than cash bonuses, incentives could take the form of accelerated promotion, prestigious assignments, and formal recognition tied to measurable outcomes such as time to deployment, number of qualified entrants, and cost reductions.
In this sense, the Pentagon requires an institutional analog to the deal-facilitating role played by investment bankers in private markets: actors who reduce information frictions, structure credible commitments, and are judged on portfolio performance rather than the success or failure of any single bet. Entrepreneurship plays a central role in many aspects of the modern economy. Firms such as Amazon, Apple, and Uber saw market openings and took risks to create products that transformed their industries.
At a time when China’s military clout continues to increase and many observers marvel at Beijing’s state capitalism model, it is time to unleash the unique competitive ecosystem of the US economy to upgrade the quality of American military power.
