Why the Recent Department of Defense JADC2 Exercises are a BFD
In the Department of Defense, Joint All Domain Command and Control (JADC2) is fundamentally about getting big systems (planes, satellites, etc.) to talk to each other and other mission systems.
DoD’s increasing investment in JADC2 programs is needed because software, data, and cross-platform integration can’t be an afterthought in big defense programs. The F35 is not just a plane; it’s a data center with a jet strapped on it that flies in the sky. The software and sensors on an F35 need to be seamlessly integrated into a broader networked system, which will not be built by the same company that builds the jet.
Recent JADC2 exercises, at Northern Command and the Army XVIII Airborne Corps, are a big freaking deal. Why? Because they show that DoD is starting to realize that AI doesn’t mean much if it isn’t put in the hands of operational commanders.
As I said on Government Matters this week, these are promising, early demonstrations of the JADC2 concept. Also true: we still have a long way to go across the Department of Defense when it comes to seamless data flow across military platforms and services.
Some background: these recent exercises were the operational testing of Project Maven's capabilities. Maven, the 4-year-old data management and computer vision program, is a collection of over 60 leading commercial cloud, software, and AI vendors integrated to make sense of the explosion of data that a new generation of satellites and other sensors are collecting. Think Google Maps for Defense, so that commanders can make better decisions in the field.
Maven, originally stood up by LT General (ret.) Jack Shanahan, and more recently run by Col Drew Cukor, is the DoD’s flagship AI program, alongside the Joint AI Center (JAIC), currently run by LT Gen Michael Groen and CTO Nand Mulchandani. Together, Maven and the JAIC are the DoD’s highest-profile investments in AI.
At Northern Command, Commander General Glen VanHerck was able to use Maven to demonstrate cross-COCOM data-enabled mission command. (Read his commentary on War on the Rocks about the importance of using data and software faster, to break down data stovepipes.)
LT General Erik Kurilla, at the Army XVIII Airborne Corps, similarly used Maven on 7,000 kilometers of satellite imagery during a live-fire exercise called Scarlet Dragon IV. “The Scarlet Dragon series is designed to increase our joint warfighting capability and how AI-augmented decision making significantly increases the scale, speed, and accuracy of our targeting process,” Kurilla told the Army Times. “It’s about seeking ways to achieve decision advantage in large-scale combat operations. It’s about learning, as an organization, how to employ data as a strategic asset in the joint fight.”
Two operational commanders (Kurilla and Maj General CD Donahue, Commanding General of the 82nd Airborne Division) were the first to buy into the power of Maven while they were in Afghanistan and now, in their new positions, they are pushing their staff and subordinate units to get on board. And not in an uncritical way either: I’m told they are the first to point out where Maven falls short. Critical (and continual) feedback from actual users of capability is key to making it better.
Most importantly, they are the type of leaders who are also forcing their units to start thinking about how to design new operational concepts based on capabilities that exist today and that might be available a year, two years from now. This is one of the first real examples, I’m told, where the buy-in from operational general officers is being matched with bottom-up innovation.
How end-users actually use a product is far more relevant than a product’s “speeds and feeds”. Product features don’t mean anything if they aren’t used. We understand this inherently with consumer products, like smartphones, some of which initially had fewer features. Apple’s first iPhone, in 2007, for example, didn’t have the tactile keyboard that the then-popular Blackberry had, and yet, it was easier-to-use—and could be used in all kinds of new ways, with new apps that were rapidly developed.
The UK Government Digital Service had business cards that had the slogan: “what is the user need”. For DoD, we need to keep asking, “what does the warfighter need?”
But too often, in DoD, the focus is on the development of the capability—which typically takes far too long to get to the field for commanders to test and use. These processes take years: writing requirements, procuring, building custom, testing, and then finally, in year 10, getting to a user with an outdated monolithic product that is so clearly not going to work for DoD in our modern software and AI age. It’s refreshing to see that parts of the DoD are starting to recognize this, slowly, but meaningfully.
Waiting around for someone to hand you the solution in this era of disruptive technologies is a recipe for failure. China is not going to slow down. The real innovation probably won't come from inside the walls of the Pentagon. It will come from the operational users who will wring out the capabilities and begin to design new ways of using the capabilities that we just could never predict from the bowels of the five-sided building. As one retired general officer told me, we have to be willing to take risks and try something really new and different. As an example, during the Battle of Britain, the UK creatively combined a network of nascent radar technology, visual observers, fighter aircraft, and a command and control system to disrupt Nazi Germany’s conventional plan to soften Great Britain. Or to use a more recent example, the US military’s airborne electronic attack crews took their conventional radar and communications jamming equipment and adjusted their tactics to combat the IED threat.
My fear—and that of many others, whose opinion and service I deeply respect—is that we won't figure out how to get innovation right, absent a conflict where we learn the hard way.
For JADC2 to work, the data has to be interoperable, machine-readable, and internally accessible. Deputy Secretary of Defense Secretary Kathleen Hicks and DoD Chief Data Officer (CDO) Dave Spirk are squarely focused on this; Deputy Secretary Hicks released a set of 5 Data Decrees, including the direction for the Department to move to an open architecture.
I also said on Government Matters that an open architecture matters because it means that DoD won’t be trapped into a vertically integrated solution. Even the large defense primes are recognizing that open standards are necessary for JADC2 to work. DoD needs to continue to push the conventional defense primes, not just on open standards, but also on true system interoperability.
This past June, Deputy Secretary Hicks launched the Acceleration of Data and AI (ADA) Initiative. She has directed the extension of Project Maven, DoD’s enterprise analytics platform, ADVANA, and the JAIC to all 11 Combatant Commands, and has created operational data teams to work closely with each command.
Hicks understands that innovation is about how data is actually used in the field. Her memo states “..the Department will seek to scale existing capabilities that have proven themselves in the battlespace and in real-world operations, simulations, experiments, and demonstrations.”
This is an important focus on getting smarter on what warfighters actually need, in part by getting new capabilities in front of users, to generate faster feedback loops. My hope is that we’ll see some real progress as operators and analysts begin inventing more warfighting concepts and begin understanding new ways they can take advantage of game-changing data-driven technologies.
Leading with Commercial
The DoD, via the JAIC, Maven, and other joint investments, is looking for ways to 'stitch the seams together' between the various Service JADC2 projects (Project Convergence for Army, ABMS for the Air Force, and Project Overmatch for the Navy). Both Maven and JAIC are looking for ways to use AI-enabled technologies to accelerate sensor-to-shooter timelines and improve decision-making.
Maven and the JAIC are also built around a "commercial first" philosophy. Here is why that is important:
The Defense Industrial Base (DIB) is made up of companies that make great planes, ships, and tanks, plus some service companies and solution integrators. But those are not software companies, nor do they know how to build modern software for DoD. Software can’t be an afterthought to big planes and ships.
We need to bring in the top software engineering talent in the country and free them to build the best solutions for DoD, rather than to “requirements” specified by industrial or service companies.
For JADC2 to really take off, a new Defense Digital Base will have to be powered by commercial “software primes” with the talent, fresh ideas, and financial resources to do the bold work required to truly operate differently.
That is why Insight Partners recently invested in Rebellion Defense, where I joined the board of directors. Rebellion is a next-generation, software and AI company focused on national defense, co-founded by Chris Lynch, former Director of the Defense Digital Service (DDS).
As Chris told NextGov, “The transformation that's happening now, going from an industrial era of defense into a software era—the world where military superiority will be defined by the flawless execution of software and technology—it's going to take a different profile of people partnering with the people who are doing the mission.”
I agree. This is the time to be bold. This is an era where software superiority will determine who has the national security advantage.
We need to get great software companies—like Rebellion Defense and others—into the game to build side-by-side with the traditional defense industrial base companies to establish the means of connecting platforms and sensors.
We also need to shift how we think about continuous improvement in software-centric platforms for defense. We need to end the concept of sustainment on JADC2 for software. Sustainment makes no sense in this context. It seems like the Air Force is coming to this recognition too. JADC2 needs to continuously evolve because competitors’ tactics change, and technology advances too rapidly, to be mired in long refresh or modernization cycles.
Insight Partners is one of the leading investors in software-as-a-service (SaaS) companies—which means our portfolio companies’ products get better over time. Every digital product in your personal life gains new capabilities continuously too. Defense needs to think in the same way.
The AI of tomorrow is built on the data of today. Making your data work for you—across systems, services, and domains—is crucial to building the AI future not of 2050, but of 2021.