With the dawn of a new year, SpaceX is resolving to restart its commercial space launches. After a pause of four months to investigate the explosion of one of its rockets, the commercial launch company announced plans to resume launches on January 8 (pushed back to the 14th due to weather). Last September’s accident sparked concerns about the safety of using commercial launch companies. With a new administration coming into office this month, and its signaled interest in space policy, it is important to highlight how essential a robust commercial space launch industry is to American interests.

The explosion of a SpaceX rocket in September led to questions from lawmakers and a NASA advisory group about the safety of using the company to launch national security satellites or astronauts. While SpaceX has said it has identified the cause of the explosion, and has worked to fix it, concerns about how to balance innovation and reliability remain. These concerns are important, given what could be at stake.

The U.S. military and intelligence communities rely heavily on satellites, the largest of which can cost over $1 billion per. Reliability is even more important when launching astronauts. This is why large national security launches still use United Launch Alliance vehicles—under a more tightly controlled contract with much higher costs—and why American astronauts still ride on Russian launch vehicles to the International Space Station. But the United States is in a high-risk transition period when it comes to space launches, regardless of how it currently awards contracts. For the best outcomes, both in terms of reliability and in costs, American policymakers need commercial companies like SpaceX to succeed.

The current position of American launch capabilities is untenable. While significant progress has been made by new entrants, they are not yet in a position to guarantee launches in the same way that the incumbent (United Launch Alliance) can. At the same time, the legacy systems that ULA uses are poised to disappear. One variant, the Delta IV, is commercially uncompetitive and ULA wants to retire it. The Atlas 5 uses Russian engines and has come under fire from Congress with the worsening of relations with Russia. ULA is working on a replacement launch vehicle, the Vulcan, that it hopes to have fully ready by 2023. The government could simply pay to keep the legacy systems, but that would only be viable for a limited time. Eventually, either due to a need for newer technology or to reduce costs, the workhorse launch vehicles currently relied on by America’s national security sector will disappear.

How, then, can the replacements best be managed? Even if ULA’s Vulcan launch vehicle arrives on time and with no issues, which cannot be guaranteed, it will only replace one retired launch vehicle. The government has a mandate to have two launch vehicles, however, to prevent a disruption to one completely shutting down launches.

So what about the second system? That system will have to come from another commercial company, all of which will come with risk. Because this risk cannot be completely mitigated during the transition to new launch vehicles, the best approach would be to encourage the commercial launch industry to produce as many potential systems as possible.

But, as the explosions of SpaceX and Orbital ATK rockets demonstrate, it can be hard when launch vehicles will have problems. The best way to develop reliability is to test the launch vehicles in real-life scenarios. Unmanned commercial and civil launches provide these scenarios. While failures in these launches have costs, they are not to the same degree as a national security or manned launch. That is exactly what is happening with SpaceX. But launching things into space is never easy, and the challenges SpaceX has faced will help it improve its launch vehicle.

Any launch vehicle that follows this path—commercial launches as tests—is going to be more risky than launch vehicles that focus first on hitting government reliability standards. This is because the design and initial funding process will have to focus on commercially competitive launch prices. In such a competition, companies are incentivized to innovate, which naturally increases the odds that things may go wrong, at least at first. Over time, however, the companies that cannot fix problems will go out of business and the rest will move towards greater reliability. So, while SpaceX’s rocket explosion was problematic, it was far from and indictment of commercial launch systems. Achieving a robust commercial launch market depends on whether these companies can locate and fix the problems, and then get back to building up a reliable track record.

The new entrants to the launch market are just that—new. But by the early 2020s, it’s expected that all of the launch vehicles used will be new. Innovation in space launches is literally rocket science. While the United States should be encouraging commercial launch companies to reach for reliability, it also needs to allow them the time to get there. America’s assured access to space depends on it.