Rebooting the Arsenal of Democracy: Anduril Mission Document
Visit rebootingthearsenal.com.
Table of Contents
0. What We Believe
1. Introduction
2. How Did We Get Here?
3. The Defense Industry Today
4. The Rise of Software
5. A New Model
6. The Government Response
7. Call to Action
0 — What We Believe
Only superior military technology can credibly deter war.
Since World War II, America and its allies’ lead
in military technology has been the pivotal factor in preventing World War III.
Today, that technological lead is in jeopardy.
The incumbent defense companies are unable to build the technology we need to reaffirm our technological lead.
We need a new breed of defense technology companies to reboot the arsenal of democracy.
1 — Introduction
If you have followed world events for the last decade, you have felt something changing. The 20th century is well behind us, and with it the sense that America and its allied democracies are the undisputed leaders of the world. World War II — the last major war that America and its allies can truly be said to have won — feels awfully long ago.
The most glaring herald of our decline is in our military’s technology. Once upon a time, the armed services of the United States, United Kingdom, and their partners across the world were envied for their science fiction technology. Today, they are equipped with tools decades behind what you or I use every day. Software has, to paraphrase Marc Andreessen, tried to eat the world, but with the battlefield it found a meal too ungainly to digest.
Despite spending more money than ever on defense, our military technology stays the same. There is more AI in a Tesla than in any U.S. military vehicle; better computer vision in your Snapchat app than in any system the Department of Defense owns; and, until 2019, the United States’ nuclear arsenal operated off floppy disks.
The absence of a crisis made us complacent. In the aftermath of the Cold War, America and its allies felt invincible. Former Soviet states appeared to be liberalizing. China’s economy was growing, and most modernization theorists believed that the Chinese Communist Party would move inexorably toward open democracy. It was the end of history — the triumph of Western ideals and a new age of American-led world order.
Until it wasn’t. Instead of acquiescing, our competitors seized the moment. China and Russia spent two decades harnessing our most powerful weapon — innovation — and built advanced weapon systems designed to neutralize and surpass our own. The results are sobering: today, in almost every wargame the United States Department of Defense models against China, China wins.¹
All too quickly, we have reached the point where, in the words of General C.Q. Brown, “We must accelerate change, or lose.”
There is a better way. Throughout World War II and the decades that followed, the West attracted the world’s most brilliant scientists and engineers, many of whom made foundational contributions to national defense. John von Neumann, Alan Turing, Kelly Johnson, the Jasons, and other gifted patriots recognized that our prosperity is only as strong as our means to defend it. Their technological innovations didn’t just bolster our military — many of their advances percolated into the consumer market. War research and development turned futuristic dreams into household staples: personal computing, GPS, the Internet, commercial air travel, and so much more.
It is time to recapture that spirit of innovation and bring cutting-edge technology to our armed forces again. War in Eastern Europe, the looming threat of a Taiwan invasion, and a rising tide of security threats in other regions of the world demand it. This is not a matter of minor reform. It will require a major re-evaluation of what kinds of technology we build, how we build it, and the speed at which we move.
Why can’t the existing defense companies simply do better? The largest defense contractors are staffed with patriots who nevertheless do not have the software expertise or business model to build the technology we need. Tomorrow’s weapons — autonomous systems, cyberweapons and defenses, networked systems, and more — are enabled through software, while these companies specialize in hardware. These companies work slowly, while the best engineers relish working at speed. And the software engineering talent who can build faster than our adversaries resides in the commercial sector, not at large defense primes. These companies built the tools that kept us safe in the past, but they are not the future of our defense.
We at Anduril are building defense technology in a new way, but we are just one company. If the allied democracies are to break free from the morass of bureaucracy, embedded habits, and mediocre technology that have failed our men and women in uniform for decades, we will need action at scale. Not just a handful, but dozens of new, innovative companies will be required to modernize our military. Tens of thou- sands of engineers will have to ask themselves if there is more to their careers than money alone. And our officials in government, without whom this effort will be for naught, will have to listen and to lead.
Some observers have suggested that young engineers do not wish to work on national security. We disagree — Silicon Valley was founded by patriots pushing science and engineering forward in the national interest and is filled with people eager to do the same again today. Others claim that it is impossible to compete with the defense giants. But the government has enthusiastically embraced nontraditional companies that get results like SpaceX and Palantir. And finally, the most pessimistic critics believe our effort to be futile because democratic leadership is in inexorable decline, as our adversaries hope it to be. We reject this. With the right technology, the right ways of building, and the right partners in the armed forces, we can build the systems that will safeguard the allied democracies for the next century and beyond.
2 — How Did We Get Here?
In his eponymous memoir about the group, former Lockheed Skunk Works director Ben Rich tells two stories. One, for which he is best remembered, is a tale of aspiration and ingenuity. Alongside his iconoclastic predecessor, Kelly Johnson, Rich and his team of “skunks” designed and built vehicles that changed how the United States waged war and gathered intelligence: the U-2 spy plane, the SR-71 Blackbird, and the F-117 Nighthawk. Lockheed’s incredible record of innovation in the national interest is defense industry mythology; Rich’s memoir is practically mandated reading for patriotic engineers.
The second story Rich tells, however, is less optimistic. Writing in 1994, Rich warns of a creeping decline among the major defense companies. “In my forty years at Lockheed,” he wrote, “I worked on twenty-seven different airplanes. Today’s young engineer will be lucky to build even one.”² Not only were new things rarely being built, Rich observed, but the costs of even trying to do so had skyrocketed. “The development costs of fighters have increased by a factor of 100 since the 1950s, and unit procurement costs have risen 11 percent every year since 1963,” Rich despaired. “Small wonder, then, that there were only seven new airplanes in the 1980s, compared to forty-nine in the 1950s.”³
Rich’s first story is history; his second is yet to conclude. The United States has not fielded a new bomber plane since the end of the Cold War. The F-35, one of the few new major systems the military is fielding, has become a byword for extravagance, currently estimated to cost taxpayers $1.6 trillion. And instead of attracting new entrants to fill the gaps, the Pentagon lost a staggering 20,500 suppliers from 2000 to 2018.⁴
When former Lockheed CEO Norm Augustine wrote that “In the year 2054, the entire defense budget will purchase just one aircraft,” he was joking — but not by much.
How did we get here? Understanding our present circumstances means understanding the history that preceded them. We do not have to delve far into the past — World War II and the years immediately following were a golden age of military innovation.
The Golden Years
“It was the civilian scientists, not the military engineers, who had been the technological innovators during the war. ‘They are the ones who made the breakthroughs,’ [General Arnold] said. He predicted that those breakthroughs — in radar, in jet propulsion, in rocketry, in nuclear weapons — would prove to be the catalysts for further innovation that would radically alter the nature of war. The First World War had been decided by brawn, he said, the Second by logistics. ‘The Third World War will be different. It will be won by brains.’”
–Neil Sheehan, A Fiery Peace in a Cold War
It can be difficult to imagine the Department of Defense, which today employs nearly 3 million Americans, moving quickly. But last century, it innovated at a speed that puts modern Silicon Valley startups to shame: the Pentagon was built in only 16 months (1941–1943), the Manhattan Project ran for just over 3 years (1942–1946), and the Apollo Program put a man on the moon in under a decade (1961–1969). In the 1950s alone, the United States built five generations of fighter jets, three generations of manned bombers, two classes of aircraft carriers, submarine-launched ballistic missiles, and nuclear-powered attack submarines.
Perhaps it should not surprise us that the government worked at the pace of a Silicon Valley startup, given that World War II and Cold War military research and development kickstarted Silicon Valley in the first place. Fred Terman, the “father of Silicon Valley,” had left Stanford in World War II to run the Harvard Radio Research Laboratory, the Department of Defense’s premiere electronic warfare lab. When he returned to Stanford after the war, he brought a wealth of government connections and influence with him. By 1947, the DoD provided half of Stanford School of Engineering’s budget. Federal government spending on science and engineering, much of it concentrated in Silicon Valley, continued to balloon — in 1960, the DoD accounted for 36% of all research and development in the world.⁶ ⁷
The DoD’s investments didn’t just benefit the military. Decades of military research sowed the seeds of prosperity in the consumer and enterprise sectors. The French media mogul and politician Jean- Jacques Servan-Schreiber, writing in 1967, marveled at the United States’ pace of innovation, predicting that “In thirty years America will be a post-industrial society… There will be only four work days a week of seven hours per day.”⁸
The commercial sector continued to advance — perhaps not as quickly as Servan-Schreiber imagined, but quickly enough for America to remain the most innovative country in the world.
But, starting in the 1960s and intensifying through the 70s and 80s, the pace of military innovation began to slow, while costs grew. In 1955, it had become clear that the pace of defense spending was unsustainable: the federal government was spending more on defense than everything else combined.⁹ Congress knew something had to change. Robert S. McNamara, a wunderkind World War II veteran with a sterling reputation for administrative efficiency at Ford Motor Company, was the man for the job.
A New Regime
In the early 1960s, under the direction of then Secretary of Defense McNamara, the Department of Defense instituted a labyrinth of new rules for acquiring military systems. McNamara was not merely reducing spending: his experience at Ford led him to believe that comprehensive reform was needed to reshape how the government bought technology. He revamped acquisitions to emphasize efficiency, the elimination of waste, and predictability. The Planning, Programming, and Budgeting System (PPBS, later changed to PPBE for “Execution”) process that he implemented is arguably the single most influential defense reform ever enacted, indelibly altering the incentives and business models of the major defense contractors.
Competition with the Soviet Union shaped McNamara’s actions — but perhaps not as modern observers would expect. Whereas today senior DoD leaders have rightly sounded the alarm on China’s ability to compete technologically with the West, McNamara did not believe the Soviet Union capable of out-innovating the United States. In the account of Brigadier General James M. Roherty, McNamara viewed the competition against the Soviet Union in terms of “quantifiables,” believing that “a relative stability characterized weapons technology in the 1960s… that ‘technological surprise’ [was] not a threat to national security and that major breakthroughs [were] not to be expected.”¹⁰
The primary focus of McNamara’s reforms was hence not innovation but cost control. He believed that, by and large, the United States already had the technology it needed to win the Cold War, and that the only problem was making sure that we didn’t spend ourselves into oblivion. The headline-grabbing arms races against the Soviet Union — the “missile gap,” the “bomber gap” — reflected differences in quantity, not quality, of military systems.
McNamara’s reforms were not wrong: they were built for a specific era with specific characteristics. Those characteristics included the federal government conducting the vast majority of research and development, rather than private companies; the overwhelming importance of large, capital-intensive, “exquisite” military systems like tanks and battleships; a predictability in the adversary’s (the Soviet Union’s) rate of technological progress; and the existence of relatively few companies able to muster the capital and expertise to build military technology.
All of those things have, to varying degrees, changed, but our system for buying technology has not. As a result, the large defense companies have learned to operate in highly idiosyncratic ways, unlike major companies in any other industry:
1. Adherence to a Lengthy Bureaucratic Process
Under the PPBS system, before the government even considers purchasing a new military system, it embarks upon a years-long process of defining requirements, deciding where and how to allocate resources, and finally releasing an award for a new system. It is extremely difficult for the Department of Defense to rapidly acquire new technology, and as a result defense companies face no pressure to develop new systems quickly. Technology developed in the commercial sector hence takes years or decades to end up on the battlefield, if it does at all.
2. Working off Onerous System Specifications
Reflecting the prevailing belief that industry was re- quired to execute, not innovate, to beat the Soviet Union, the requirements for major military programs are spelled out in extensive detail. Unlike most industries, which are driven forward by the creativity of the most successful companies, defense contractors are rarely asked to find creative ways to solve problems, and are sometimes punished for doing so.
3. Spending Little on Internal Research & Development
Because McNamara’s reforms made it exceptionally difficult to buy new technology quickly or to buy technology for which there is no defined requirement, defense companies rarely develop products of their own accord. Typically, they sell existing systems (which was most of their business under McNamara) or they wait for the government to order specific R&D efforts, for which they are directly compensated. For a comparison, the largest technology companies today — whose revenues vastly exceed those of the largest defense companies — spend roughly 10–20% of their revenue on research and development. Newer or mid-sized technology startups might spend closer to 60% or 70%. The major defense companies spend 1–4%.¹¹
4. Prioritizing Proposals Over Performance
McNamara believed duplicative programs to be at the root of DoD waste and would not tolerate government spending on multiple development efforts for similar new systems, whether between military services or among different vendors. This means that once a company is awarded a major contract, it is extremely difficult to take it from them. The incentive for large defense firms is hence to spend heavily on teams of lawyers and lobbyists to shape program requirements in line with the company’s existing technology. This political battle becomes just as important as building the product itself. These lobbyists are often former military officials: By 1969, over 2,000 military officers per year were leaving the DoD to work for a major defense contractor, three times as many as in 1959.¹³
5. Tolerating Prolonged Failure
In most industries, a company that fails to produce a functional product goes out of business. In the defense industry, when a company is three, five, ten years into a program and has failed to build what they promised, the government is stuck between a rock and a hard place — do they cancel the contract and throw away years of development, potentially bankrupting the vendor in the process? Or do they begrudgingly provide the vendor with even more money to salvage the program? Typically, it’s the latter. In the words of the infamous city planner Robert Moses, “Once you sink that first stake, they’ll never make you pull it up.”¹⁴
This may simply be a poor decision (sunk cost fallacy), but sometimes it is part of a deliberate policy to pre- serve the “defense industrial base” — a critical mass of defense companies that can be called upon in an emergency. There is value to this policy, but it has significant costs. It creates moral hazard for major defense companies, which effectively become “too important to fail,” and ensures that not all contracts are given out meritocratically, blunting companies’ incentives to produce better products.
To reiterate, there were highly logical rationales for each of these rules and norms. Some of McNamara’s beliefs turned out to be wrong, but others were vindicated — the West won the Cold War, after all. One thing McNamara’s DoD got emphatically right was reducing the use of cost-plus-fixed-fee (CPFF) contracts from 38% of contracts awarded to just over nine percent (though by 2020 they had crept back up to comprise 38% of DoD’s contracts once again).¹⁵ ¹⁶ Cost-plus contracts pay vendors for the costs of developing and producing new systems, plus a percentage fee on top (typically 6–8%). This means that working quickly and efficiently often conflicts with contractors’ profit motives. In the words of Lt. Gen. James Stansberry, “Unless changes are made in the current profit system that demands higher costs as a prerequisite for higher profits, it is futile to expect lower costs… there is little economic motivation for contractors to reduce direct or indirect costs.”¹⁷
Ultimately, however, McNamara’s system was a bet that the United States could predict the future — that they had a good idea of how Soviet technological development would progress and that, with careful planning, the United States could remain ahead. That is simply no longer true. Our adversaries innovate incredibly quickly, partially because they have learned from the mistakes of the Soviets, partially because they have become incredibly adept at industrial espionage,¹⁸ and partially because software, unlike hardware, can be built and improved upon incredibly quickly (more on this below).
Six decades after they were first introduced, the rules for doing business with the Pentagon remain, in the words of former Defense Secretary Donald Rumsfeld, “one of the world’s last bastions of central planning.”¹⁹ The primary aim of these rules — to keep costs in check — was a failure: in 1969, Robert Benson, an analyst in the Office of the Secretary of Defense, testified that “about 90% of major weapon systems that the Defense Department procures end up costing at least twice as much as was originally estimated.”²⁰ The incentives these rules created for defense companies have ossified into these companies’ very ways of being. It is not malice or incompetence that prevents the large defense companies from innovating, or moving quickly, or investing in internal research and development — it is simply a rational response to the incentives with which they are presented.
3 — The Defense Industry Today
One remarkable statistic tells a damning story about the modern defense sector: Since 1963, the defense industry has never suffered a negative ten-year period.
Is this because defense companies have built an unprecedented succession of valuable products? Far from it. On a long enough time horizon, defense companies are simply not punished for failure (and, indeed, the horizon is not all that long). Individual companies may receive a slap on the wrist for particularly egregious missteps, but by and large the government accepts that, frustrating as it may be, slow timelines, minimal innovation, and swelling costs are “the way things are.”
Part of the reason that the large defense companies keep getting away with underperformance is that they face little competition. In 1993, the Cold War fading into memory, then Deputy Secretary of Defense Bill Perry sat down with the CEOs of the major defense firms at what came to be known as “the last supper.”²² Perry warned them that substantial cuts to the defense budget were coming and that many of their companies would not be able to survive. Put simply, he said, “We expect defense companies to go out of business. We will stand by and watch it happen.” This precipitated a flurry of mergers and acquisitions.
Perry attempted to counterbalance the impact of consolidation by lowering the barriers to entry into the defense industry. He discouraged the use of onerous military specifications (“milspecs”) for new systems except as a “last resort”²⁴ and championed 1994’s Federal Acquisitions Streamlining Act (FASA), which mandated the use of commercial, off-the-shelf alternatives over bespoke new technology. He also encouraged program managers in the DoD to “take more risks” and bet on new technology and new companies. Many of these changes, however, were not enforced. Both SpaceX and Palantir were forced to sue their largest customers for a failure to adhere to Perry’s rules,²⁵ ²⁶ while a RAND corporation study from the early 2000s found that program managers had neither the incentives nor capacity to make meaningful bets on new companies and technology.²⁷
The result is an ageing, top-heavy industry that moves slowly because nobody is chasing it. The ten largest defense companies, all of which were founded decades ago, account for upwards of 80% of the industry’s revenue. Nearly two thirds of major weapons systems contracts in the United States have just one bidder.²⁸ Shareholders buy defense stocks for their stability, their predictability. These companies move slowly and break as little as possible.
Most engineers, understandably, do not wish to work on national security under these conditions. Engineers want to see their code deployed, their robots in motion, their products out in the world making an impact. Working slowly on antiquated technology that may never be deployed is far from an engineer’s dream. And so the companies once staffed with some of the greatest engineering minds in the Western world now struggle to attract graduates from top engineering programs.
4 — The Rise of Software
The Department of Defense used to own all the technology it needed to credibly deter and, if necessary, win the next great war. But the same is not true today.
This isn’t because our ships aren’t big enough, or our planes fast enough, or our tanks sturdy enough. The areas in which the United States and its allies are falling behind are not these large hardware systems, which we employed to dominate the 20th century, but sophisticated software. The majority of the Department of Defense’s innovation priorities — including artificial intelligence, networked weapons, cybersecurity, and more — are enabled, either in part or in total, by software. The hardware components of swarming munitions, autonomous UAVs, or cruise missile defense systems are just one half of the equation.
Software will change how war is waged. The battlefield of the future will teem with artificially intelligent, unmanned systems, which fight, gather reconnaissance data, and communicate at breathtaking speeds. The bipartisan congressional Future of Defense Task Force, drawing upon DARPA’s “Mosaic Warfare” concept, describe a future “where ubiquitous and affordable unmanned air and ground platforms find targets on a contested battlefield and pass the information to a decision maker who can instantly task another part of the same system to strike the enemy from safety.”²⁹ The traditional division between land, air, sea, and space domains will become increasingly anachronistic, as nation states compete for total, all-domain superiority.
There is sometimes a misconception among government officials that software is easier to build than hardware, or that higher quality human capital is required to produce hardware than software. This is flatly untrue: Building the software that makes these concepts possible is an engineering task on the order of magnitude of building a fighter jet. It will require the work of the very brightest engineers in the allied democracies.
Silicon Valley has a near-monopsony on this talent. Almost every single top-tier software engineer in the democratic world works at a technology company, whether at a startup or in Big Tech. John von Neumann, whose talents were so highly valued that the RAND corporation paid him to scribble down his morning thoughts while shaving,³⁰ would today be working for Google or Facebook. Everything about how such companies operate has been tailored to these engineers — the creative and professional freedom, the speed and lack of bureaucracy, even the casual dress code. There is also an economic incentive for these engineers to work at technology companies, which have higher margins than most other industries and can afford to pay their top talent more.
The most unrealistic part about James Bond’s gadgets isn’t the technology itself — it’s that the government owns it. Indeed, the companies building transformative software technology rarely work on national security at all. Sometimes, this is for cultural reasons — after employees at Google protested Google’s participation in Project Maven, the DoD’s flagship AI program, Google canceled their contract. Google isn’t alone; the largest tech companies either avoid defense work or treat it as a secondary priority. Moreover, the defense technology they do build is rarely designed for the job. More often, it is commercial software adapted to a military purpose. These companies deserve our applause for working with the government, but dual-use technology won’t be enough to maintain our technological lead.
In fairness to the tech employees who’d rather not work on national security, they didn’t sign up to work on defense, and there is no reason in a liberal democracy that one should be forced to work on things that one finds objectionable. But it is striking how far behind the times Silicon Valley has been on questions of national security. The tech industry prides itself on being the first to spot new trends. On national security and the rise of our strategic adversaries, it was the last.
The traditional defense contractors, meanwhile, struggle with software not only because they lose engineering talent to Silicon Valley, but because modern software development practices conflict with their very ways of being. The legacy primes are used to slowly developing large, exquisite hardware systems like battleships, which sit in a factory for years before they are finally complete. Once sold, they are largely finished, besides occasional maintenance. Software is fundamentally different. It is developed by shipping a minimum viable product out the door as quickly as possible and seeing where it fails. Software companies respond rapidly and continuously to the performance of software in the field, to such an extent that deployment and iteration are parts of the development process itself. As software technology advances, developers add new features and keep a product up to date.
That is what the military systems and military companies of today should look like: more Tesla than Ford, more Apple than Nokia.
5 — A New Model
The biggest lie of the political debate surrounding defense spending is that we are stuck in a dichotomy between doing “more with more” or “less with less.” This is a false choice — technology means doing more with less. The stagnation in the defense industry is not inevitable, and we need not accept it.
Anduril is one of a slew of innovative defense companies pioneering a new model, based on the following principles:
1. Outpacing the Threat
Security threats are evolving faster than we are. In 2018, Under Secretary for Research and Engineering Mike Griffin noted that, on average, it takes the US six- teen years to deliver an idea to operational capability, compared to under seven years for China.³¹ That kind of advantage compounds incredibly quickly: an adversary that starts out with only a fraction of our capabilities but which advances 2x faster will quickly overpower us. And it isn’t just great powers moving more quickly than we are — in the Middle East, well-financed terrorist organizations and militias iterate monthly on their armed drones.³² In the cyber domain, a greater number of state and nonstate actors alike have become capable of sophisticated attacks.
We need defense companies that are fast. That won’t happen simply by wishing it to be so: it will only happen if companies are incentivized to move. Thankfully, there is a readily available model for this: every single other industry on the planet. Whether they are selling cars, phones, furniture, or coffee, most companies do not get paid until they build a finished product. That can and should be the case in the defense industry, too: companies should privately finance their research and development, engineer new products quickly and efficiently, and then sell them “off the shelf” to their government partners. The government, in turn, should appropriately reward companies that take on risk and development costs for the technology they build.
There is recent and compelling precedent for this: SpaceX. Despite working on capital-intensive, highly complex products, SpaceX has poured billions of dollars of private capital into research and development. It was found by NASA’s own analysis to be three times cheaper than NASA itself.³³ If rocket-ships can be built quickly and sold “off the shelf,” then so can many more things in the defense budget.
But the threat doesn’t stop changing once we sell a system, so nor should our technology. It beggars belief that our men and women in uniform receive updates on their smart phones whenever technology improves, but must wait years for updates to the technology that might save their lives. Software is a double-edged sword: updates can be distributed to operational theaters across the globe with the click of a button, which means that software-enabled technology can stay up to date nearly indefinitely. Conversely, if software isn’t updated frequently, it quickly becomes obsolete. The government should demand the continuous deployment of the latest and greatest software to the tactical edge.
2. Building to the Mission, Not to Spec
The defining feature of history’s most successful technology companies is product vision. Jeff Bezos, Steve Jobs, Elon Musk, and other great technology entrepreneurs all had distinct images of the future and how their products would shape it. These visions were independent of — and often directly contradicted by — the opinions of those around them. They worked on things that their customers and competition did not believe were possible or did not yet understand. Had they been building new products off painstaking specifications laid out by their customers, the history of modern technology would look very different.
The next great defense companies will also have a vision of the future: they will build to the mission, not to spec. That means designing products to solve problems from first principles. Defense acquisitions have historically been dominated by an emphasis on exquisite platforms — i.e., the Navy needs a fleet with X ships, or the Air Force needs Y planes — whose capability requirements are laid out in excruciating detail. This process answers the question of what the next battleship should look like, but never the question of whether a new battleship is needed at all.
Today’s model has things backwards. The problems we are trying to solve should be top of mind, not only for the government, but for the companies building the solutions to those problems. Nobody understands the operational threats that the government is engaging with better than the government itself, but, in turn, nobody understands the technical solutions that can be brought to bear better than industry. Not every company’s product vision will ultimately meet the operational needs of their partners in government, but that is okay — most companies in a capitalist system fail. Competition will bring the best to the forefront and solve problems in ways the government had not predicted or thought possible.
3. Software-First
Software is finally eating the battlefield, whether the defense industry likes it or not. Whether it is AI-enabled drone swarms, networked weapons systems, real-time situational awareness generated by sensor fusion, or cyberwarfare, software is at the core of the weapon systems of the future. Building world-class software platforms is an engineering challenge akin to building a fighter jet: the next generation of defense companies will invest in doing so accordingly.
Pure software can radically expand the capabilities of an armed force, but it can become even more potent when integrated into hardware systems. Take unmanned aerial systems, for example. UAVs like General Atomics’ Reapers and Predators have been staples of the United States’ and United Kingdom’s arsenals for decades now, but because these vehicles are manually piloted and monitored, they require a crew of a dozen to operate. Cutting-edge artificial intelligence and autonomy can reverse that equation.
This is but one example. The broader point is that modern technology companies do not think of the sheet metal on the exterior of a system as its most valuable element: they instead build the hardware around the new capabilities that software enables. For almost any major commercial technology from the last de- cade — new phones, cars, voice assistants, hobby drones — everything from sensor selection to control surfaces is optimized for software.
4. Controlling Defense Budgets
The defense of the nation is the pre-condition for the nation’s prosperity — but defense companies have a duty, like any entity that receives public funding, to deliver services to the government at a minimal burden to the taxpayer.
This is a particularly sensitive issue for the defense industry. Governments around the world routinely (and justifiably) complain of being bilked³⁴ by major defense contractors.³⁵ While there are certainly historical examples of fraud and abuse on the part of contractors, the deeper problem is structural. High costs are endemic to a defense industry that operates off cost-plus contracts.
On the other hand, the uncomfortable truth is that it is impossible to build and scale a successful business without rewarding the investors and employees who took a risk on it. If the companies challenging the incumbent defense firms are not profitable, they will get nowhere at all, and our defense technology will continue to languish behind consumer tech.
There is no magic bullet for driving down costs, but a healthy dose of free market capitalism would go a long way in getting us there. The more companies we bring into the defense industrial base, the more these companies will have to compete on price. The more that companies spend on their own dime rather than relying on cost-plus contracts, the more they’ll cut costs. And the more we leverage new technology like AI, the more we can save on labor costs for the dull, dirty, dangerous jobs of military service.
6 — The Government Response
With the principles above, we can create a new breed of modern, agile defense companies that turn our soldiers into superheroes. These companies will solve problems of their own accord, but also bring out the best in the incumbent defense contractors, who have not felt competition nipping at their heels in decades.
But the companies following these principles, however well-intentioned, will mostly fail without an equivalent response from our government leaders. The major defense companies operate as they do because our laws and regulations incentivize it. Governments across the democratic world have a monumental task ahead of them in reshaping the incentives of the defense industry and allowing a more dynamic, innovative model to thrive.
We have spent the last few years publishing and advocating a series of policy analyses and recommendations to modernize our antiquated weapons acquisitions system, usually tailored to the United States government but often applicable across nations. Recommendations like these are just the tip of the iceberg, and we exhort our leaders in government to develop and drive forward solutions where they see problems. A few areas we have identified where we can make major progress include:
This isn’t just for companies: the government must embrace a software-first mindset too. In practice, this would involve a long series of changes, including but not limited to:
a. Establishing large, critical programs for key software priorities.
Simply put, the government does not value software in the way that it does hardware. In the United States, fewer than 10% of the military’s Major Defense Acquisition Programs (MDAPs) are software systems. Even more damningly, the military’s largest software programs are mostly IT tools, designed for internal management and civil functions, most of which could be done better by commercially available technology. The first step toward leveraging the latest in modern technology is demanding it through big programs.
b. Awarding prime contracts to software companies.
Fundamental design decisions for new military systems are typically made before software companies are even consulted, meaning the hardware is not optimized for and often entirely unable to accommodate cutting-edge software.
c. Aligning acquisition policy with industry best practices.
This would include transitioning away from RDT&E- heavy, cost-plus fixed fee contracts towards firm fixed price contracts that encourage rapid prototyping, deployment, and testing of new capabilities. To go even further, governments should experiment with industry-standard payment structures like subscriptions and “as-a-service” models, which minimize switching and sunk costs, encourage competition, and demand that technology is constantly updated.
2. Running Meritocratic Competitions for New Systems
Most government contracts are awarded on the basis of lengthy proposals, with little to no reference to real-life performance of the technology in question. To encourage the most innovative companies and engineers to work on defense technology, the government should instead:
a. Shift from proposals-focused competition to performance-based competition.
Instead of prioritizing proposals, acquisitions should be structured as a series of bake-offs and well delineated projects that test a vendor’s ability to solve a problem with technology and deploy it operationally.
b. Always have a meaningful contract at the end of a competition, and issue it quickly.
Innovation grants and pilot contracts only go so far — in return for investing time and money into demonstrating a new capability, companies that successfully solve government problems should be rewarded in a timely manner.
c. Keep rewarding new winners by frequently recompeting large programs.
Governments routinely award decade-long, multibillion dollar contracts to single vendors. This couldn’t be more different to the commercial world, where consumers are typically free to switch between vendors at minimal cost.
d. Measure outputs, not inputs.
Governments should hold themselves accountable: instead of asking how many competitions they held, they should ask themselves how many of those competitions led to a meaningful contract for a nontraditional defense company.
3. Revamping Anachronistic Data Rights Practices
The government typically demands “data rights” — or technical information about how to build a given piece of technology — in a sweeping, one-size-fits-all fashion, a relic of a hardware-centric world. This, unfortunately, scares away all but a handful of modern software companies, who view giving the government their valuable source code as an existential threat to the company. A new model of data rights agreements would:
a. Account for the size of the program in question, allowing vendors to retain greater rights for small programs and giving the government greater control for larger programs.
b. Closely define the scope of the use of the software being acquired.
c. Address technical lock-in by demanding adherence to technical interoperability standards such as APIs and Government-defined or industry-defined open protocols.
d. Allow vendors to propose alternative contract structures, including as-a-service contracts.
e. Weigh the costs, not just the benefits, of demanding greater data rights.
4. Helping Companies Cross the ‘Valley of Death’ by Loosening Contracting Timelines
In the United States, it is easier than ever to win a small pilot or prototype contract from the government thanks to a smorgasbord of new innovation entities like the Defense Innovation Unit, AFWERX, NavalX, and more. Conversely, it has never been harder to transition to a large program of record, leaving companies languishing for years between initial funding and scaled deployments. This often sends companies bankrupt, or, more typically, dissuades companies from ever doing business with the government in the first place. This year, the United States Congress addressed this problem with an “Agile Procurement Transition Pilot,” designed to help scale new technologies quickly and get companies through the valley of death. This is a valuable start, but by no means a wholesale solution to the valley of death. Governments across the democratic world should find ways to speed up their acquisitions processes and quickly get new technology into the hands of their uniformed men and women.
5. Leveraging Existing Programs
Often, the government has created programs to access and scale new innovations. Because the process to acquire new technologies is so Byzantine, however, acquisitions officers are often unaware of or under-educated about the acquisition tools at their disposal. Where possible, governments should publicize and educate acquisition and contract officers about the tools they already possess to minimize bureaucracy and access new technology.
Since founding Anduril, we have been profoundly encouraged by the volume and intensity of support ideas like these have received from those in government. There is a growing group of government officials in administrations across the world who recognize that business as usual simply will not cut it anymore. The challenge ahead is gigantic, but so are the rewards of success: continued peace and prosperity in the democratic world.
7 — Call to Action
There is no secret government silo of advanced technology that will save us if war breaks out — you must build it. If you have read this far, you care deeply about the future of our collective defense. As engineers, government leaders, and citizens:
Help us to rebuild the arsenal of democracy and make that future safe, prosperous, and free.
References
¹ Brose, Christian. The Kill Chain. 2020. p. xii.
² Rich, Ben. Skunk Works: A Personal Memoir of My Years at Lockheed. 1994. p. 316.
³ Rich, p. 321.
⁴ Werner, Ben. “New Pentagon Report Points to Problems in the U.S. Shipbuilding Industrial Base.” USNI News, October 2018, https://news.usni.org/2018/10/05/u-s-shipbuilding-example-of-defense-industrial-base-weakness.
⁵ Chart taken from Greenwalt, William & Patt, Dan, “Competing in Time: Ensuring Capability Advantage and Mission Success through Adaptable Resource Allocation.” Hudson Institute, February 2021, https://www.hudson.org/research/16717-competing-in-time-ensuring-capability-advantage-and-mission-success-through-adaptable-resource-allocation.
⁶ If you are interested in learning more about the history of early Silicon Valley and its overlap with the U.S. military, we recommend Steve Blank’s “The Secret History of Silicon Valley”: https://steveblank.com/secret-history/.
⁷ Congressional Research Service, “The Global Research and Development Landscape and Implications for the Department of Defense.” 2021, https://sgp.fas.org/crs/natsec/R45403.pdf.
⁸ Servan-Schriber, Jean Jacques, 1967. The American Challenge.
⁹ Ronald Fox, J. Defense Acquisition Reform, 1960–2009: An Elusive Goal. 2011, https://www.hbs.edu/ris/Publication%20Files/11-120.pdf, p. 2.
¹⁰ Roherty, James Michael. Decisions of Robert S. McNamara: A Study of the Role of the Secretary of Defense. 1970. p. 106.
¹¹ Maucione, Scott. “DoD’s Kendall wants more research spending from industry.” Federal News Network, November 2015, https://federalnewsnetwork.com/defense/2015/11/kendall-wants-research-spending-industry/.
¹² Jaruzelski, Barry; Chwalik, Robert; & Goehle, Brad. “What the Top Innovators Get Right.” Strategy + Business, October 2018, https://www.strategy-business.com/feature/What-the-Top-Innovators-Get-Right
¹³ Hartung, William D. Prophets of War: Lockheed Martin and the Making of the Military-Industrial Complex. 2010. p. 103.
¹⁴ Goldberger, Paul. “Robert Moses, Master Builder, Is Dead at 92.” New York Times, July 1981, https://www.nytimes.com/1981/07/30/obituaries/robert-moses-master-builder-is-dead-at-92.html.
¹⁵ Lofgren, Eric. “A History of Thought in Defense Acquisition.” International Cost Estimating and Analysis Association, June 2017. http://www.iceaaonline.com/ready/wp-content/uploads/2017/07/PS09-Paper-Lofgren-History-of-Thought-in-Defense-Acquisitions.pdf.
¹⁶ Lofgren, Eric. “Did SpaceX and Palantir transform government contracting? A VC’s view.” Acquisition Talk, August 2021, https://acquisitiontalk.com/2021/08/did-spacex-and-palantir-transform-government-contracting-a-vcs-view/.
¹⁷ Testimony of Lt. Gen. James Stansberry, United States Air Force (Ret.) (formerly commander, Electronics Systems Division, Air Force Systems Command), before the Senate Subcommittee on Defense Acquisition Policy of the Committee on Armed Services, “Defense Procurement Process,” February 1985, p. 20.
¹⁸ Lucas, Ryan. “DOJ’s China Initiative aims to counter theft of U.S. secrets and technology.” NPR, November 2021, https://www.npr.org/2021/11/04/1052248809/dojs-china-initiative-aims-to-counter-theft-of-u-s-secrets-and-technology?t=1639407638565.
¹⁹ Rumsfeld, Donald. Public Statements of Donald H. Rumsfeld, Secretary of Defense, 2001. 2001. https://books.google.com/books/about/Public_Statements_of_Donald_H_Rumsfeld_S.html?id=hGFHAQAAIAAJ. p. 1728.
²⁰ Fox, p. 40.
²¹ Hamtil, Lawrence. “Exploring the Surprising Resilience of the Defense Industry.” Fortune Financial, February 2021, https://fortunefinancialadvisors.com/blog/exploring-the-surprising-resilience-of-the-defense-industry/.
²² Mintz, John. “How a Dinner Led to a Feeding Frenzy.” The Washington Post, July 1997. https://www.washingtonpost.com/archive/business/1997/07/04/how-a-dinner-led-to-a-feeding-frenzy/13961ba2-5908-4992-8335-c3c087cdebc6/.
²³ Berenson, Doug; Higgins, Chris; & Tinsley, Jim. “The U.S. Defense Industry In a New Era.” War on the Rocks, January 2021. https://warontherocks.com/2021/01/the-u-s-defense-industry-in-a-new-era/.
²⁴ Fox, p. 172.
²⁵ Judson, J. “Judge Rules in Favor of Palantir in Lawsuit Against US Army.” Defense News, October 2016, https://www.defensenews.com/land/2016/10/31/judge-rules-in-favor-of-palantir-in-lawsuit-against-us-army/.
²⁶ Davenport, Christian & Fung, Brian. “Elon Musk’s SpaceX to sue government over space launch contract.” The Washington PostApril, 2014. https://www.washingtonpost.com/business/economy/elon-musks-spacex-to-sue-government-over-space-launch-contract/2014/04/25/1001aa6e-cca6-11e3-95f7-7ecdde72d2ea_story.html.
²⁷ Fox, p. 182.
²⁸ “Courage to Learn: Defense & Aerospace.” American Economic Liberties Project, January 2021, https://www.economicliberties.us/our-work/courage-to-learn-defense-aerospace/.
²⁹ Future of Defense Task Force, “Report 2020.” September 2020, https://armedservices.house.gov/2020/9/future-of-defense-task-force-releases-final-report, p. 73.
³⁰ Jacobsen, Annie. The Pentagon’s Brain: An Uncensored History of DARPA, America’s Top-Secret Military Research Agency. 2015. p. 30.
³¹ Greenwalt, p. 35.
³² Arraf, Jane & Schmitt, Eric. “Iran’s Proxies in Iraq Threaten U.S. With More Sophisticated Weapons.” New York Times, June 2021, https://www.nytimes.com/2021/06/04/world/middleeast/iran-drones-iraq.html.
³³ Chaikin, Andrew. “Is SpaceX Changing the Rocket Equation?” Smithsonian Magazine, January 2012, https://www.smithsonianmag.com/air-space-magazine/is-spacex-changing-the-rocket-equation-132285884/.
³⁴ V. Roberts, Steven. “Congress: The Provocative Saga of the $400 Hammer.” The New York Times, June 1984, https://www.nytimes.com/1984/06/13/us/congress-the-provacative-saga-of-the-400-hammer.html.
³⁵ NYT staff. “Lockheed Is Accused of Waste and Deceit on C-5A.” The New York Times, September 1971, https://www.nytimes.com/1971/09/30/archives/lockheed-is-accused-of-waste-and-deceit-on-c5a-lockheed-deceit-on.html.
