The world is on the cusp of a new era of warfare dominated by unmanned systems, artificial intelligence, networked weapons and sensor fusion. We recently got a glimpse of this future in the rapid devastation that Turkish-backed Azeri forces delivered to the Armenian military with low-cost drones and smart weapons. While China and Russia exploit advances in emerging technologies, the United States military struggles to adopt them. If the U.S. does not adapt quickly to this new way of fighting wars, we will become vulnerable to attacks from authoritarian regimes that pose an existential threat to our national security, our democracy and our way of life.

Donald Trump ran for president as a disruptor, and one area he promised to disrupt was national defense. But while speaking the language of disruption, Trump spent even more on the same old ships, vehicles and aircraft at the expense of emerging technology. He got historic increases in defense spending from Congress and squandered them on the past.

President-elect Joe Biden and his team must do what Trump promised and failed to do: Disrupt national defense. To his credit, Biden has acknowledged that the United States’ decline in military technology is a threat to our national security. But while recognition is a crucial first step, the Department of Defense needs a new innovation playbook. As Air Force Chief of Staff Ge. C.Q. Brown said: “We must accelerate change, or lose.”

We recommend the Biden administration pursue the following initiatives to stay ahead of China and defend American values at home and abroad:

Think software first: As the congressional Future of Defense Task Force has concluded, the DoD will need to engage nontraditional defense companies to build the “essential technology” that will maintain the United States’ military advantage. The DoD must understand these technology systems are valuable because of the complex software they run on, not the hardware casing on the outside. Unfortunately, the DoD remains mired in old habits, attempting to acquire these leap-ahead capabilities by making software subservient to hardware and subordinating AI, autonomy and software-to-hardware manufacturers.

In order to recruit software companies to build the next generation of defense systems, the DoD needs a cultural shift on its approach to software. DoD leaders must understand that the engineering challenges of building a fighter jet and those of building AI-powered systems are of equivalent magnitude and should be treated with equivalent respect. And they must start awarding major defense programs to nontraditional defense contractors, who can draw upon the world’s most talented software engineers in ways the traditional defense base cannot.

The DoD should also revisit its approach to software data rights (the term for license rights in government contracts). The DoD’s insistence on owning source code and reserving the right to pass that code from one vendor to another is a nonstarter for commercial companies. It means that they are neither guaranteed any meaningful recurring revenue nor are they able to effectively sell their products to multiple customers, destroying their incentives to sell to the military. The DoD’s concerns about vendor lock-in can be addressed in less obtrusive ways, such as requiring that software companies building their government software be “open” — interoperable, extensible and adaptable to solutions from industry.

Pick winners: The DoD is paralyzed by a cultural and bureaucratic aversion to “picking winners” — worrying that by handing out contracts, it is engaging in favoritism and anti-capitalist behavior. But nothing could be more anti-capitalist than failing to reward the best talent and ideas.

We cannot choose winners blindly, of course. The DoD should hold more frequent demonstrations and competitions to assess who is building the best technology. However, once the best technologies have proven themselves, the DoD should spend large sums scaling these technologies into programs of record.

After picking winners, the DoD should loosen its financial sustainment requirements for some systems. Requiring decades-long sustainment plans made sense during the Cold War: Our adversaries were predictable, meaning our technology was usually a step ahead, and many of the most important systems we procured were clearly intended to be operational and relevant for decades. Neither is true today: China is a dynamic adversary constantly innovating, and important tech like software and attritable aerial systems improves every year.

We used to be better at this. The CIA and the DoD “picked a winner” when they determined that Kelly Johnson’s team at Skunk Works was the only one able to build revolutionary stealth technology like the U-2 spy plane or F-117 Nighthawk. The DoD picked winners when it identified the Massachusetts Institute of Technology’s Radiation Laboratory and Harvard’s Radio Laboratory as hotbeds of innovation. Across our history, identifying the foremost experts in emerging fields, and giving them the resources to innovate, has yielded the United States’ greatest-ever military advances.

Introduce a scaling innovation fund: Finally, the DoD needs a quicker and simpler pathway to scale emerging technology into widespread deployments.

Nontraditional, venture-backed defense companies work on timelines of 12-18 months between investments. During such periods, they must show meaningful revenue increases, or risk being denied new funding. But the DoD allots contracts painfully slowly; the program objective memorandum process, the mechanism for apportioning defense funding, begins almost three years in advance. This means, at best, our military technology is 3 years old, and too often that commercial innovation never makes its way to the military.

One remedy would be for Congress to appropriate a “Scaling Innovation for the Warfighter” fund to provide funding to transition capabilities from pilot programs to programs of record in a relevant time frame. This would tackle two problems: providing war fighters gamechanging new technology in real time, and closing the two-year “valley of death” between pilot and program of record that kills so many new companies.

The Biden administration has a chance to make history by reinvigorating innovation in national defense and ensuring our military maintains its leadership. The administration also could simply ignore the problem or conclude that it is too large and cumbersome to fix. China and Russia are counting on President-elect Biden to choose the second option; we urge him to choose the first.

Trae Stephens is a partner with Founders Fund and is the chairman of Anduril Industries. Steve Blank is an adjunct professor with Stanford University. He recently resigned from the Defense Business Board.