Reboot the Defense Innovation Ecosystem
The venture capital model of investment is predicated on a high failure rate, justified by massive returns on a small number of successes. Yet prestigious, venture-capital-backed defense tech startups are failing more than even this model can tolerate. The overwhelming majority of defense tech startups are terminally stuck in the Valley of Death, and venture capital bets on defense tech startups aren’t paying off. Soon, the best venture capital firms will stop investing in the federal market, and the newly rebuilt bridges between Silicon Valley and DC will collapse.
There has been only one major initial public offering (IPO) in the defense market’s recent past — Palantir in 2020 — and that’s a hugely negative signal to venture capital firms about the value of the market. As the head of federal deployments at Scale, I’ve witnessed this dynamic from a front-row seat. Scale is an artificial intelligence startup with both market-leading commercial technologies and production-level federal government contracts. Unfortunately for the startup industry and the U.S. Department of Defense, that makes Scale the exception, not the rule.
While we can’t predict the future, asymmetric technology capabilities (especially those using machine learning) will certainly challenge our traditional defense advantages across domains. The defense community must treat innovation seriously to build the most contingency-ready defense industrial base possible. Katherine Boyle from Andreessen Horowitz neatly summarized the key issues in a tweet storm. She’s right — venture capital firms aren’t seeing the returns on investment to justify bets even on dual-use technologies in the public sector, let alone pure defense tech startups, and time is running out.
Because they are pre-IPO, defense tech startups don’t report their financials publicly. This makes it impossible to holistically assess the health of the defense tech startup ecosystem. But I speak with defense tech stakeholders every day — from founders and CEOs to business development managers and engineers. Though these are limited and anecdotal data points, I see that companies have four primary paths: become a Palantir, exit by acquisition, leave the public sector market, or languish as a zombie company. Let’s review each of these:
So far, there is only one Palantir, and Palantir’s success as a pathfinder launched a thousand defense tech ships. None of those newer entrants have gone public, although SpaceX and Dataminr are close. Palantir’s path to IPO was famously protracted, even for a company that promised to help the U.S. government mitigate intelligence failures in an exploding post-9/11 defense market. Despite those strong tailwinds, Palantir had to fight their own customers for years (including suing the Army) to win key public sector market segments and finally IPO.
What about an exit by acquisition? This can work for defense tech startups with a product that fills a specific gap in the portfolio of a much larger business. Successful exit by acquisition (usually dual-use, not pure defense tech companies) examples include Wickr (now AWS), Expanse (now Palo Alto Networks), and the public sector side of Hivemapper (now Palantir). While sometimes acceptable to venture capital firms depending on the deal terms, acquisitions are usually not an optimal outcome.
Relatedly, a company can leave the public sector market quietly to double down on commercial work. If you ask the CEOs of these businesses, they’ll say they’re still in the public sector — though they’ve dramatically reduced or even eliminated their federal teams. Many tech startups interested in the defense market win Small Business Innovation Research work via defense innovation framework entities like the Defense Innovation Unit, SOFWERX, or AFWERX and then issue excited press releases about entering the defense market. AFWERX alone awarded 1,436 pilot contracts in FY20. Pilot wins feel real to startups, and so they invest in building defense-specific teams and technologies. As their pilots move to conclusion, the reality sets in that pilots rarely transition to production-level contracts, and boards and investors apply pressure on the company’s executive team to cut their losses and focus on commercial work.
Then there is death by zombification. A company can simply just hang around until someone stops paying the web hosting fees. I don’t know exactly how many companies on the Dcode website are zombies, but I would bet the number is substantial. These are defense tech startups that are out of investment runway (i.e., they’re broke). I could name a few confirmed zombies here, but that would damage any chance they have of ever coming back to life. This is where most defense tech startups are headed, and that would be acceptable to venture capital firms if there were only more successes.
I concur with Josh Wolfe of Lux Capital that the problem is grave, but I disagree that we need “to get the big ‘primes’ — as the country’s leading defense contractors are known — out of the way.” America needs Lockheed Martin’s aircraft factories, Huntington Ingalls’ shipyards, and the other primes because national security requires many forms of resilience. As an example, Raytheon produces the Javelin used to great effect in Ukraine today. It’s also unlikely that venture capital firms will back a startup to produce critical but controversial technologies such as intercontinental ballistic missiles. However, large primes have trouble attracting the best talent. Few software engineers who graduate from top-tier computer science programs want to work for the large primes. Software engineers prefer startups, where they have access to the latest technology, can build or grow their own teams, and are able to gain valuable equity in the business.
Partnerships between the traditional primes and tech startups are vital and are the first of four key ways to reboot the Department of Defense’s approach to innovation. This is not a comprehensive set of recommendations, which would fill many pages, but it is a summary of the most important.
The four approaches needed to successfully reboot:
- Incentivize and enforce the commercial item preference for software at the subcontract level.
Robust, high technology readiness-level commercial capabilities exist for nearly the full range of the Department of Defense’s technology requirements, from damage assessments to autonomy. These commercial technologies are not reaching the Department of Defense, since large defense primes traditionally operate those programs. Large primes should be rewarded for effectively integrating commercial technologies — and taking on the associated compliance challenges — in the contract award decision evaluation factors (per the Federal Acquisition Regulations part 15.304).
- Protect contracting officers and program managers and reward them for taking smart risks.
We don’t need to wait for the findings of the Commission on Planning, Programming, Budgeting, and Execution Reform to begin fixing this problem — we must introduce incentives and protections for contracting officials who are warranted with the stewardship of taxpayer dollars now. This is non-controversial, and Heidi Shyu, the undersecretary for research and engineering at the Defense Department, has publicly and repeatedly made this same point. Contracting officers and program managers are rarely rewarded for creative thinking in fostering a robust defense technology ecosystem — but they are certainly punished in the form of limited career trajectories if even a smart bet doesn’t pan out.
- Streamline the Defense Counterintelligence and Security Agency facility clearance process.
The Defense Counterintelligence and Security Agency’s mission is vital to the nation. They are charged with overseeing the National Industrial Security Program to ensure that the defense industry properly safeguards the classified information in their possession while working for the U.S. government. Yet, the process to obtain a facility clearance — basically a security clearance for a company — to perform classified work undermines that mission: it makes our country less safe because the facility clearance process acts as a moat keeping out the companies who can bring best-in-class technologies to bear for defense. The facility clearance process can take years to complete depending on a startup’s organizational structure and investors, and it imposes onerous compliance and reporting burdens on very lean teams. The Department of Defense needs American innovation inside its classified networks and should find ways to bring non-traditional defense contractors into that ecosystem.
- Pick defense tech startup production contract winners on a regular basis.
Through both the Federal Acquisition Regulations and mechanisms like Other Transaction Authorities, the U.S. government has the ability to award large production contracts to tech startups. Anduril recently won a $1 billion-ceiling counter-unmanned aerial systems contract with the U.S. Special Operations Command. This is an example of what right looks like — but a healthy defense tech startup ecosystem requires ten or more of these types of awards per fiscal year. Consistent, predictable, annually recurring revenue from the U.S. government helps give companies the solid financials they need to IPO.
Critical voices on the Hill, like Rep. Ken Calvert, understand that the defense tech ecosystem is failing. Rep. Calvert’s $100 million warfighter innovation fund is a critical step in rebooting that innovation ecosystem. Still, the Department of Defense needs to do more, and Congress should be a strategic investor and partner with the defense innovation sector.
Competition with China is not the only risk to U.S. national security. Unforeseeable threats will emerge as global dynamics evolve, and new technologies erode the gap between commercial solutions and national technical means. The Department of Defense needs America’s leading technologists on its side. It’s not a luxury or theater, but a necessity to maintain defense agility. Warfighters, analysts, and decision-makers require fielded, cutting-edge capabilities now. They don’t need one Palantir, they need 30 Palantirs working in partnership with them to shape the future of U.S. national security.
Shands Pickett leads federal deployments for Scale AI, Inc. (a late-stage AI company and a top 3 global data startup worth $7.3 billion). Previously, he created and led the national security business development team for another startup, Premise Data Corp. He served on two theater deployments to Afghanistan as part of the U.S. Army Human Terrain System.