Why Trump’s ‘Golden Dome’ Won’t Shield the U.S. from Nuclear Strikes

During a briefing from the Oval Office this week, President Donald Trump revealed his administration’s plan for “Golden Dome”—an ambitious high-tech system meant to shield the U.S. from ballistic, cruise and hypersonic missile attacks launched by foreign adversaries. Flanked by senior officials, including Secretary of Defense Pete Hegseth and the project’s newly selected leader, Gen. Michael Guetlein of the U.S. Space Force, Trump announced that Golden Dome will be completed within three years at a cost of $175 billion.

The program, which was among Trump’s campaign promises, derives its name from the Iron Dome missile defense system of Israel—a nation that’s geographically 400 times smaller than the U.S. Protecting the vastness of the U.S. demands very different capabilities than those of Iron Dome, which has successfully shot down rockets and missiles using ground-based interceptors. Most notably, Trump’s Golden Dome would need to expand into space—making it a successor to the Strategic Defense Initiative (SDI) pursued by the Reagan administration in the 1980s. Better known by the mocking nickname “Star Wars,” SDI sought to neutralize the threat from the Soviet Union’s nuclear-warhead-tipped intercontinental ballistic missiles by using space-based interceptors that could shoot them down midflight. But fearsome technical challenges kept SDI from getting anywhere close to that goal, despite tens of billions of dollars of federal expenditures.

“We will truly be completing the job that President Reagan started 40 years ago, forever ending the missile threat to the American homeland,” Trump said during the briefing. Although the announcement was short on technical details, Trump also said Golden Dome “will deploy next-generation technologies across the land, sea and space, including space-based sensors and interceptors.” The program, which Guetlein has compared to the scale of the Manhattan Project in past remarks, has been allotted $25 billion in a Republican spending bill that has yet to pass in Congress. But Golden Dome may ultimately cost much more than Trump’s staggering $175-billion sum. An independent assessment by the Congressional Budget Office estimates its price tag could be as high as $542 billion, and the program has drawn domestic and international outcries that it risks sparking a new, globe-destabilizing arms race and weaponizing Earth’s fragile orbital environment.

On supporting science journalism

If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

To get a better sense of what’s at stake—and whether Golden Dome has a better chance of success than its failed forebears—Scientific American spoke with Jeffrey Lewis, an expert on the geopolitics of nuclear weaponry at the James Martin Center for Nonproliferation Studies at the Middlebury Institute of International Studies.

[An edited transcript of the interview follows.]

It’s been a while, but when last I checked, most experts considered this sort of plan a nonstarter because the U.S. is simply too big of a target. Has something changed?

Well, yes and no. The killer argument against space-based interceptors in the 1980s was that it would take thousands of them, and there was just no way to put up that many satellites. Today that’s no longer true. SpaceX alone has put up more than 7,000 Starlink satellites. Launch costs are much cheaper now, and there are more launch vehicles available. So, for the first time, you can say, “Oh, well, I could have a 7,000-satellite constellation. Do I want to do that?” Whereas, when the Reagan administration was talking about this, it was just la-la land.

But let’s be clear: this does not solve all the other problems with the general idea—or the Golden Dome version in particular.

What are some of those other problems?

Just talking about space-based interceptors, there are a couple [of issues that] my colleagues and I have pointed out. We ran some numbers using the old SDI-era calculation from [SDI physicists] Ed Teller and Greg Canavan—so we couldn’t be accused of using some hippie version of the calculation, right? And what this and other independent assessments show is that the number of interceptors you need is super-duper sensitive to lots of things. For instance, it’s not like this is a “one satellite to one missile” situation—because the physics demands that these satellites … have to be in low-Earth orbit, and that means they’re going to be constantly moving over different parts of the planet.

So if you want to defend against just one missile, you still need a whole constellation. And if you want to defend against two missiles, then you basically need twice as many interceptors, and so on.

You probably have to shoot down missiles during the boost phase, when the warheads are still attached. For SDI, the U.S. was dealing with Soviet liquid-fueled missiles that would boost, or burn, for about four minutes. Well, modern ones burn for less than three—that’s a whole minute that you no longer have. This is actually much worse than it sounds because you’re probably unable to shoot for the first minute or so. Even with modern detectors [that are] much better than [those] we had in the 1980s, you may not see the missile until it rises above the clouds. And once it does, your sensors, your computers, still have to say, “Aha! That is a missile!” And then you have to ensure that you’re not shooting down some ordinary space launch—so the system says, “I see a missile. May I shoot at it, please?” And someone or something has to give the go-ahead. So let’s just say you’ll have a good minute to shoot it down; this means your space-based interceptor has to be right there, ready to go, right? But by the time you’re getting permission to shoot, the satellite that was overhead to do that is now too far away, and so the next satellite has to be coming there. This scales up really, really fast.

Presumably artificial intelligence and other technologies could be leveraged to make that sort of command and control more agile and responsive. But clearly there are still limits here—AI can’t be some sort of panacea.

Sure, that’s right. But technological progress overall hasn’t made the threat environment better. Instead it’s gotten much worse.

Let’s get back to the sheer physics-induced numbers for a moment, which AI can’t really do much about. That daunting scaling I mentioned also depends on the quality of your interceptors, your kill vehicles—which, by the way, are still going to be grotesquely expensive even if launch costs are low. If your interceptors can rapidly accelerate to eight or 10 kilometers per second (km/s), your constellation can be smaller. If they only reach 4 km/s, your constellation has to be huge.

The point is: any claim that you can do this with relatively low numbers—let’s say 2,000 interceptors—assumes a series of improbable miracles occurring in quick succession to deliver the very best outcome that could possibly happen. So it’s not going to happen that way, even if, in principle, it could.

So you’re telling me there’s a chance! No, seriously, I see what you mean. The arguments in favor of this working seem rather contrived. No system is perfect, and just one missile getting through can still have catastrophic results. And we haven’t even talked about adversarial countermeasures yet.

There’s a joke that’s sometimes made about this: “We play chess, and they don’t move their pieces.” That seems to be the operative assumption here: that other nations will sit idly by as we build a complex, vulnerable system to nullify any strategic nuclear capability they have. And of course, it’s not valid at all. Why do you think the Chinese are building massive fields of missile silos? It’s to counteract or overwhelm this sort of thing. Why do you think the Russians are making moves to put a nuclear weapon in orbit? It’s to mass kill any satellite constellation that would shoot down their missiles.

Golden Dome proponents may say, “Oh, we’ll shoot that down, too, before it goes off.” Well, good luck. You put a high-yield nuclear weapon on a booster, and the split second it gets above the clouds, sure, you might see it—but now it sees you, too, before you can shoot. All it has to do at that point is detonate to blow a giant hole in your defenses, and that’s game over. And by the way, this rosy scenario assumes your adversaries don’t interfere with all your satellites passing over their territory in peacetime. We know that won’t be the case—they’ll light them up with sensor-dazzling lasers, at minimum!

You’ve compared any feasible space-based system to Starlink and noted that, similar to Starlink, these interceptors will need to be in low-Earth orbit. That means their orbits will rapidly decay from atmospheric drag, so just like Starlink’s satellites, they’d need to be constantly replaced, too, right?

Ha, yes, that’s right. With Starlink, you’re looking at a three-to-five-year life cycle, which means annually replacing one third to one fifth of a constellation.So let’s say Golden Dome is 10,000 satellites; this would mean the best-case scenario is that you’re replacing 2,000 per year. Now, let’s just go along with what the Trump administration is saying, that they can get these things really cheap. I’m going to guess a “really cheap” mass-produced kill vehicle would still run you $20 million a pop, easily. Just multiply $20 million by 2,000, and your answer is $40 billion. So under these assumptions, we’d be spending $40 billion per year just to maintain the constellation. That’s not even factoring in operations.

And that’s not to mention associated indirect costs from potentially nasty effects on the upper atmosphere and the orbital environment from all the launches and reentries.

That, yes—among many other costly things.

I have to ask: If fundamental physics makes this extremely expensive idea blatantly incapable of delivering on its promises, what’s really going on when the U.S. president and the secretary of defense announce their intention to pump $175 billion into it for a three-year crash program? Some critics claim this kind of thing is really about transferring taxpayer dollars to a few big aerospace companies and other defense contractors.

Well, I wouldn’t say it’s quite that simple.

Ballistic missile defense is incredibly appealing to some people for reasons besides money. In technical terms, it’s an elegant solution to the problem of nuclear annihilation—even though it’s not really feasible. For some people, it’s just cool, right? And at a deeper level, many people just don’t like the concept of deterrence—mutual assured destruction and all that—because, remember, the status quo is this: If Russia launches 1,000 nuclear weapons at us—or 100 or 10 or even just one—then we are going to murder every single person in Russia with an immediate nuclear counterattack. That’s how deterrence works. We’re not going to wait for those missiles to land so we can count up our dead to calibrate a more nuanced response. That’s official U.S. policy, and I don’t think anyone wants it to be this way forever. But it’s arguably what’s prevented any nuclear exchange from occurring to date.

But not everyone believes in the power of deterrence, and so they’re looking for some kind of technological escape. I don’t think this fantasy is that different from Elon Musk thinking he’s going to go live on Mars when climate change ruins Earth: In both cases, instead of doing the really hard things that seem necessary to actually make this planet better, we’re talking about people who think they can just buy their way out of the problem. A lot of people—a lot of men, especially—really hate vulnerability, and this idea that you can just tech your way out of it is very appealing to them. You know, “Oh, what vulnerability? Yeah, there’s an app for that.”

You’re saying this isn’t about money?

Well, I imagine this is going to be good for at least a couple of SpaceX Falcon Heavy or Starship launches per year for Elon Musk. And you don’t have to do too many of those launches for the value proposition to work out: You build and run Starlink, you put up another constellation of space-based missile defense interceptors, and suddenly you’ve got a viable business model for these fancy huge rockets that can also take you to Mars, right?

Given your knowledge of science history—of how dispassionate physics keeps showing space-based ballistic missile defense is essentially unworkable, yet the idea just keeps coming back—how does this latest resurgence make you feel?

When I was younger, I would have been frustrated, but now I just accept human beings don’t learn. We make the same mistakes over and over again. You have to laugh at human folly because I do think most of these people are sincere, you know. They’re trying to get rich, sure, but they’re also trying to protect the country, and they’re doing it through ways they think about the world—which admittedly are stupid. But, hey, they’re trying. It’s very disappointing, but if you just laugh at them, they’re quite amusing.

I think most people would have trouble laughing about something as devastating as nuclear war—or about an ultraexpensive plan to protect against it that’s doomed to failure and could spark a new arms race.

I guess if you’re looking for a hopeful thought, it’s that we’ve tried this before, and it didn’t really work, and that’s likely to happen again.

So how do you think it will actually play out this time around?

I think this will be a gigantic waste of money that collapses under its own weight.

They’ll put up a couple of interceptors, and they’ll test those against a boosting ballistic missile, and they’ll eventually get a hit. And they’ll use that to justify putting up more, and they’ll probably even manage to make a thin constellation—with the downside, of course, being that the Russians and the Chinese and the North Koreans and everybody else will make corresponding investments in ways to kill this system.

And then it will start to really feel expensive, in part because it will be complicating and compromising things like Starlink and other commercial satellite constellations—which, I’d like to point out, are almost certainly uninsured in orbit because you can’t insure against acts of war. So think about that: if the Russians or anyone else detonate a nuclear weapon in orbit because of something like Golden Dome, Elon Musk’s entire constellation is dead, and he’s probably just out the cash.

The fact is: these days we rely on space-based assets much more than most people realize, yet Earth orbit is such a fragile environment that we could muck it up in many different ways that carry really nasty long-term consequences. I worry about that a lot. Space used to be a benign environment, even throughout the entire cold war, but having an arms race there will make it malign. So Golden Dome is probably going to make everyone’s life a little bit more dangerous—at least until we, hopefully, come to our senses and decide to try something different.