Some D-Day Thoughts On Another Invasion
The Geeks Have Established a Beachhead in the Defense Industry
We really don’t want the American defense industry to ossify.
Even someone who doesn’t follow geopolitics closely (me) can quickly put together a list of hotspots and risks faced by America and our allies. The war in Ukraine and a newly emboldened Russia. China’s intentions globally, and specifically in the South China Sea. Iran and its many proxies. North Korea’s possible nukes and missiles. Who even knows how many death cults, fanatics, and terrorists advancing the technologies and techniques of asymmetric warfare.1
I didn’t break a sweat putting that list together, but looking back over it it made me sweat a little.
Yes, we Americans are safe now. We have a formidable military (on which we spend much as the next ten countries combined), a great many nuclear warheads, and close allies on both of our borders. I’ll be beyond surprised if anyone launches an invasion against us anytime soon.
But we can all agree, can’t we, that we need to stay sharp? And that if we see evidence we’re losing our edge, we should take that evidence seriously instead of being complacent about American national security?
Getting Slower at Building Fast Planes
One of the most worrying pieces of evidence I’ve seen recently is the chart below. It shows how long it’s taken to develop successive military aircraft vs two other kinds of complicated machines: commercial aircraft and cars.
That sharp, steady upward trend in military aircraft development timelines is bad news. Is it justified by the ever-increasing complexity of those birds? Nah. Military aircraft have always incorporated the latest and greatest tech, but from the end of WWII to the mid-1970s we were able to pretty consistently develop new ones in about five years. Now it’s more like 25.
Does anyone think the V-22 or F-35, which began development around the turn of the century, are incorporating the latest sensors or machine learning models? Me neither. As startup analyst Jeff Burke puts it,
There is more AI in a Tesla than any military vehicles... Snapchat has better computer vision models than most of our military technology. The military was using floppy disks as recent as 2019.
Burke makes a key point:
“A popular culture assumption that has been depicted in Hollywood movies and shows is that the US has a secret arsenal. When calamity hits, we will pull out high tech weapons that nobody has. The truth is... we are still using many weapons that are decades old.
The battles of those eras are way different of what the next war will be. It will be driven by autonomous vehicles, software, etc. Today, we have serious structural problems.
Burke has a great post on this platform outlining these problems. I want to concentrate here on one of them — one that until recently would have seemed pretty intractable to me. Recently, though, I’ve become less worried about it, because of the healing power of competition. If that competition continues and intensifies things are going to get better in the defense industry. Either a group of Industrial Era Incumbents are going to shape up, or a bunch of modern business geeks are going to take their business from them.
A Sleepy Sector
Throughout the 21st century to date, American defense contracting has been dominated by a “big five:” Boeing, General Dynamics, Lockheed Martin, Northrup Grumman, and Raytheon. They receive more than a third of total US defense spending, but that stat presents too rosy a picture of the state of competition in the industry. As Burke writes, “In some cases, there are now only 1-2 potential US-based contractors competing to make a specific weapon on a military program.”
The big five have been around for a while. As Mario Gabriele writes in his ‘stack The Generalist, “these businesses trace their roots back to the early 20th century. General Dynamics began life as the Electric Boat Company in 1899. Northrop Grumman is the “Big Five’s” equivalent of a scrappy up-and-comer: one of its original entities started in the relative modernity of 1930 – the same year a cartoon mouse named “Mickey” made his debut in a comic strip.”
So we’ve got a defense sector dominated by a handful of veteran companies. They operate in a highly capital- and knowledge-intensive environment. Despite the frictions that crop up in any long relationship, they’re quite close to the Pentagon. The US defense industry, then, could be categorized as an oligopolistic near-monopsony with high barriers to entry. Economists believe that firms operating in such an environment have strong tendencies to ramp up prices and ramp down costly innovation. And the Big Five have acted as if their highest priority is to confirm these beliefs.
Gabriele nicely summarizes the prices-up, R&D-down state of the US defense industry at present
The lack of competition has inflated pricing. Between 2008 and 2018, the inflation-adjusted cost of Pentagon weapons systems grew by 13%. With no rivals, defense contractors are often emboldened to charge more than they meet in an efficient market. For example, in 2019, Transdigm, a large, publicly-traded contractor, was found to be marking up its products by as much as 9,400%. Transdigm charged $4,361 for a half-inch “drive pin” estimated to cost $46. The company eventually agreed to repay $16 million after a governmental investigation…
Perhaps companies whose founding occurred closer to the American Civil War than our current year should not be expected to be exemplars of innovation. Contractors ship products slowly and invest little in experimentation.
Boeing spends roughly the same amount on R&D as it did in 2001. Its total investment in 2021 totaled $2.2 billion, approximately 3.5% of its revenue. Northrop Grumman spent a similar percentage. Compare that to a company like Alphabet. The Google parent organization spent $31.5 billion or more than 12% of total revenue on R&D. Meta spent even more on a relative basis, investing $24.6 billion or 21% of revenue.
One other aspect of the US defense industry that dampens participants’ hunger to improve is the prevalence of cost-plus-fixed-fee (CPFF) contracts, which currently account for nearly 40% of those awarded by the Pentagon. Under these arrangements, the DoD essentially guarantees what the profit from the project2 will be — that’s the fixed fee — by agreeing up front to cover the contractor’s costs, however large they grow. This is good work if you can get it, and as long as your definition of “good” doesn’t include notions of efficiency or thrifty use of taxpayer dollars.3
I suggest that the environment of the big five— oligopolistic; monopson-ish; clubby; full of covered costs, guaranteed profits, ramped-down research, and incremental innovation — is an environment susceptible to missing something big in the arena of national security, like a major change in combat itself brought on by technological progress (sensors, networks, AI, automation, etc.) or a sea change in how complicated things such as modern defense systems can and should be conceived, designed, built, and operated. My extremely uninformed view is that the former is happening. My better-informed view is that the latter is absolutely taking place, as I wrote here and here and in my whole dang book The Geek Way.
I also suggest that the environment of the big five might not just miss new stuff. It might also get worse over time at doing the same old stuff. The graph above of increasing timelines for developing military aircraft illustrates this concern. I wrote about the phenomenon a bit in my previous post here.
The Elixir of Competition
The phenomenon is easy to describe and disheartening to think about: we should expect organizations to get jammed up over time — to become less able to do the thing(s) that they’re supposed to do. One of the ideas that hit me hardest as I was researching The Geek Way comes from anthropologist Joe Henrich, who writes in his fantastic book The Secret of Our Success
Over time, history suggests that all prosocial institutions age and eventually collapse at the hands of self-interest, unless they are renewed by the dynamics of intergroup competition. That is, although it may take a long time, individuals and coalitions eventually figure out how to beat or manipulate the system to their own ends, and these techniques spread and slowly corrode any prosocial effects.
Swap out “prosocial institution” for “organization” or “company,” and “any prosocial effects” for “ability to get its job done” and the conclusion still holds. The enemy is in fact us, according to Henrich. We ultrasocial humans create organizations, and then we jam them up.
I’ll write more about this jamming up later, because it’s such an important and (I believe) misunderstood phenomenon. Here I want to highlight the other fundamental human phenomenon Henrich describes: intergroup competition.
The most obvious kind of intergroup competition is head-to-head combat, which sounds bad. Wars, border skirmishes, turf battles between rival street gangs or drug cartels — these are not to be desired. But one kind of intergroup competition is: competition among rival firms for customers and markets. We cheerleaders for capitalism shake our pom poms because intense competition for people’s business has the counterintuitive effect of leaving them better off. It spurs efficiency, innovation and other good stuff far better than any other economic arrangement we’ve come up with. There are important exceptions to this rule,4 but the rule stands: contested markets are engines of betterment for people and societies.
They’re also the best way to inject some vitality into an industry in danger of getting jammed up. Which is why I’m glad that Anduril exists, and is going great guns. It and other defense tech startups have important roles to play in keeping our national defense strong.
Poking Holes in the Wall
In 1993 secretary of defense William Perry kicked off a huge wave of corporate consolidation at “The Last Supper,” a legendary meeting at which he told contractors that US national security spending was going to shrink as the Cold War faded. Perry saw that something like the big five would result, and tried to ensure that they’d have competition from younger companies. He largely failed, but Ash Carter made some headway during the Obama administration. As a remembrance for Carter, who died in 2022, put it, he
created several ways for small startups to get access to Pentagon leaders—processes he called “poking holes in the wall” since “there's no door to get into the Pentagon for companies that don't know how to work the Department of Defense.” His successors have expanded those efforts even further.
Anduril reflects the success of those efforts, as well as the dynamism of the US entrepreneurship scene. The company was founded in 2017 by five members of California’s high-tech ecosystem. It came about because Trae Stephens, an investor who had previously worked at Palantir, grew frustrated at the lack of startups working with the government on technologies related to national security.
Stephens began talking to Palmer Luckey, who had founded the virtual reality headset company Oculus and sold it to Facebook. After Luckey and Facebook parted company in 2017 he and Stephens decided to start a new company. Three former colleagues — two from Palantir, once from Oculus — joined them as founders of the new venture. Their goal was to build a variety of autonomous hardware and sensor networks, all integrated by software that made heavy use of AI.
Within five years of its founding Anduril received its first billion-dollar contract. It was from the US Special Forces Command for a variant of the Anvil, a drone designed to autonomously destroy other uncrewed aerial vehicles. The Anvil project was born during a weekend brainstorming session in early 2019. Initial prototypes showed promise, and by summer Anduril was claiming a near-perfect success rate. By the end of that year, the company was shipping the Anvil to military clients. In early 2023, according to some reports, it was being used in Ukraine.
In early 2024 Anduril was awarded major contracts to build both undersea and airborne autonomous systems for the Pentagon. The company beat out Boeing, Lockheed Martin, and Northrup Grumman to be named one of the two main suppliers to the Air Force’s Collaborative Combat Aircraft program, which is intended to develop autonomous planes that act as “wing men” for crewed fighter planes. The program plans to purchase at least 1000 of these planes at a target price of $21 million to $28 million each.
Does anyone think that Anduril is going to take anything like 25 years to deliver this fighter? Again, me neither. That question’s probably not super fair or informative, though, since autonomous aircraft are much simpler than crewed ones. So here’s another query: what is Anduril’s development timeline going to look like compared to that of General Atomics, the other company picked as a contractor to the CCA?
I know nothing about GA other than it was founded in 1955. It’s an Industrial Era incumbent, in other words, which tells me a lot about how it manages big projects. And which leads me to predict that Anduril vs. GA is going to look much like SpaceX vs. Boeing, or Tesla vs. fill-in-the-blank other American or European carmaker. I hope that I’m wrong about this, and that the intergroup competition here is an intense and close one. Not because I’m a GA shareholder (it’s privately held), but because I want our national security to improve at geek speed, not incumbent speed.
I sure hope the intelligence community how many such groups there are.
An accounting prof would read those words and admonish me that projects and products don’t have a profit or loss — only entire companies do. Individual projects and products make a contribution to that bottom line. To which I’d respond with something like thanks, nice bowtie.
An accounting prof who specializes in CPFF contracts would admonish me here that there are all kinds of provisions included in such contracts to ensure that the supplier can’t fleece the goverhment run up costs wantonly or endlessly. I’m sure there are.
Grab any good Econ 101 textbook and you’ll find that most of it is devoted to these exceptions — to market failures.