Transferring military technology takes time, teams, and money—for now.

Technology transfer and national security—everyone talks about it, and most everyone needs it. This week the Atlantic Council hosted a discussion with some European diplomats on just how to make it happen, and afterwards, we jotted down some further thoughts. Transferring or co-developing technologies effectively demands more than running blueprints through the photocopier, or cyber-stealing some source code. Assimilating know-how across national borders and commercial boundaries means cooperating with potential competitors—a challenging task for both contractors and military forces. In effect, military technology transfer and co-development today are full-contact, team sports, requiring serious managerial attention. But developments in two important technological fields could change that, and with potentially deleterious effects on the security of more technologically advanced states.

One shouldn’t assume that technology always flows between the usual suspects. For a start, consider the MRAP concept. For a long time, the Middle East didn’t build arms and armor, it bought them. Now, the UAE’s Nimr Automotive has effectively taken over the RG35 project (er, N35 project) from South Africa’s OMC. Insurgents work the issue too. Da’esh is sourcing bomb-making materials from 51 companies in 20 countries. According to report published in February by Conflict Armament Research, Mr. Baghdadi’s goons are even getting their remote detonators from a firm in Arizona. As Andrew Tilghman reported last month for Military Times, the study noted that other Iraqi militants had done so prior to 2011, which “indicates some continuity, and possibly a technology transfer, between different groups.” None of this is new, of course. As I quoted a defense industry executive back in 2008, from Sri Lanka to Iraq to Chechnya, insurgents have long had an “international business model” for technological development.

For its part, the US Defense Department is statutorily mandated to transfer what technology it can to state and local governments, and to the private sector. One might reasonably wonder how to balance the national security value of technological secrets with the economic value of that already paid-for research. As Robert Farley of the University of Kentucky wrote earlier this week, up until now, Chinese military power has depended on technology transfer—or just outright intellectual property theft. Today, he notes, almost every national defense industry depends on regular transfusions of know-how from more commercial activities. So it’s reasonable to ask just how much Chinese weapons makers needed to steal the JSF plans, and how much they could learn by sending a few more students to Berkeley, scrutinizing a Boeing jetliner, or just dissembling one of those made-in-China iPhones.

But technology transfer works both ways. In the 2000s, the federal military depots secured statutory authorities to work more closely with contractors in remanufacturing projects. What they learned greatly improved their efficiency. Through its Better Buying Power initiative, the Defense Department has also been seeking to expand competition by mandating technology transfer through the purchase of technical data rights and packages. As officialdom is sure to learn, gaining knowledge is not just a matter of photocopying the blueprints. The Army commissioned a study of this issue over 40 years ago, and concluded that while its second sources had some serious quality and schedule problems, “documentation inadequacies were not a major cause of product inadequacies or time overruns.” (See National Materials Advisory Board, The Effectiveness of the Army Technical Data Package in Technology Transfer for Procurement, NMAB-325, 1975.) A few years later, George Daly and James Schuttinga concluded that technology transfer between military contractors had historically proven costly, to the point that paybacks took a minimum of three years. (See “Price Competition and the Acquisition of Weapon Systems,” Journal of Policy Analysis and Management, 1982.) Remember: that’s when the transfer is encouraged and paid-for.

This because on some level, as I wrote in October 2013, effective technology transfer and even technological co-development depends on one’s ability to sell a solution to a customer who’s also a competitor. Consider the Gripen NG project between Sweden’s Saab and Brazil’s Embraer. As I wrote when the deal was announced in December 2013, Brazilian Defense Minister Celso Amorim said that “the choice took into account performance, technology transfer and cost, not only for acquisition but also for maintenance.” At $125 million per plane, the cost appeared higher than that rejected by Swiss voters, who were offered a price of about $110 million. These contracts vary greatly, of course, with the degree of bundled through-life support, but much of the difference may have come from the technology transfer needed to support Brazilian desire for a second production line. Brasilia appears to be getting what it needs: Saab and the Swedish government needed development partners in a way that France and the United States didn’t, so it’s likely that Stockholm permitted things that Washington and Paris wouldn’t.

The details remain undetermined, but clearly some side-by-side learning will necessary. In Sweden, Linköping University is already planning to host a small legion of Brazilian engineers. This is because technology transfer is like rugby: it demands good scrumming. As wrote recently Mike Anderson, CEO of the placement firm Elevator, “the top engineers tend to move in teams.” (Yes, that’s his business model, but bear with me.) The closeness of that cooperation then begs the question—is the Gripen NG deal foremost a way of sharing development costs, a strategic partnership for future sales, a means of technology transfer between companies and countries, or just the best way to match two ambitious, extended enterprises in an alliance to develop some excellent technology? Whatever the strategy, it’s for the long haul; those entering into a fighter jet project should anticipate that it will last for at least 25 years.

We haven’t even gotten to questions of organizational absorptive capacity. During the Cold War, the Soviets worked hard to steal technology, and in 1991, still watched their weapons and whole war-fighting system get pummeled on television. As Michael Horowitz wrote in The Diffusion of Military Power, learning how to use a ballistic missile is a lot easier than learning how to use an aircraft carrier. He was studying military operations, but something similar can be said for military technologies: the more complex, the harder to copy across organizational boundaries. Rockets and land mines, check; fixed-wing flight operations at sea, maybe not.

All that said, there remains cause for concern. If additive manufacturing proves as promising as enthusiasts assert, copying an adversary’s military technology could become as simple as cyber-stealing the 3-D printer plans. If artificial intelligence proves so useful too, using another’s technology could become as simple as cyber-stealing the tactics algorithms. Perhaps this doesn’t net as much as an aircraft carrier, or even an infantry fighting vehicle, but it might permit the easy transfer of swarming drones and mobile mines. If the scary starts looking even partly true, the sometimes casual effort paid to safeguarding leakage of intellectual property may require more focused attention.

James Hasík is a senior fellow at the Brent Scowcroft Center on International Security.