Nuclear energy seems to be back in favor this week. At the COP29 climate change conference, discussions are underway about how nuclear can help the world reduce carbon emissions.

Also, the Biden administration announced a goal to triple nuclear energy production in the U.S. by 2050.

The Inflation Reduction Act is helping with that with a ton of subsidies. And those subsidies are part of the reason that this year, Amazon, Microsoft and Google each signed massive deals to build nuclear projects.

The companies have a voracious appetite for electricity as artificial intelligence data centers chew up more and more energy.

The U.S. lost its appetite for building new nuclear power plants several decades ago for a few reasons. First, it's incredibly expensive to build.

Also, meltdowns and environmental risks loom large in the public imagination. But a new form of nuclear power generation called small modular reactors, aims to change that.

This is THE INDICATOR FROM PLANET MONEY. I'm Darian Woods. And I'm Wailin Wong.

Today on the show, could nuclear energy work differently this time? We probe the CEO of the company working with Google on its small modular reactor project, and we ask how it will deal with concerns about cost, safety and waste.

Google wants to power all its data centers without fossil fuels. But it's not just satisfied using solar, wind and batteries.

With IRA subsidies and the prospect for always-on energy, nuclear power is a tempting possibility.

In October, Google announced it would buy a ton of nuclear energy from a company called Kairos Power starting in 2030.

If this plan succeeds, it would be the equivalent of enough power for more than 400,000 homes.

Google's deal is for energy from what's called small modular reactors. They're small, as in potentially as little as a three-story building, modular, as in there'd be a lot of them made the same way.

Now, to be clear, there are no modern small modular reactors producing energy commercially in the U.S. right now.

But they are one proposed answer to a big issue with traditional nuclear power, which is that it's really expensive. Mike Laufer is the CEO of Kairos Power, the company that Google is working with.

Nuclear plants today, in the conventional way, is really expensive, and it's not obviously going to be economically better than alternative sources.

Throughout the 1950s and '60s, nuclear engineers initially tried to drive down costs by harnessing economies of scale.

So instead of building a few medium-sized reactors, you could build one really enormous one. And that did work in bringing down electricity costs in many cases.

Nowadays, running an existing nuclear facility is cost-competitive with solar and wind and natural gas.

But the costs of building, starting, and eventually decommissioning a nuclear power plant are enormous. It's far more expensive than methods like solar with battery storage.

Nuclear power plants are also susceptible to cost overruns. Among megaprojects, nuclear is among the worst offenders of cost blowouts. The average nuclear plant ends up at more than double the planned expense.

There are two major reasons for these cost overruns. First, each plant is unique - different locations, different local safety considerations, different sizes.

That makes it easier to mess something up when the new design doesn't account for all eventualities.

When you have really long, capital-intensive projects or megaprojects, there's a lot of process that gets added in there,

and there's the sense that you don't get a lot of opportunities to build the big thing, and so people tend to put more into it because you kind of want to get as much out of that opportunity as you can.

And that's tremendously expensive. The second reason is that because it takes so long to build a plant, experience is hard to find.

When those cycles are 20 to 30 years, as they are in nuclear, you need to just make sure that, like, you have processes that people remember what you did, you know, 20 years ago, which is not an easy thing.

Small nuclear modular reactors aim to address both those problems. By building small and modular, they'll hopefully be simplified and identical each time, and they'll allow more people to become experts at building them.

Actually building things and getting that experience and knowing where to invest time and energy is absolutely crucial.

And what's very elusive in nuclear technology is being able to follow that learning curve.

A learning curve, meaning basically, each time something is built, developers get experience that leads to a more efficient process.

And therefore, that thing gets cheaper to produce. Solar and wind are good examples with costs now a tiny fraction of what they were a couple of decades ago. But this has not happened with nuclear energy historically.

So that's the concept. I think that the reality of translating that idea into reality is it shouldn't be trivialized. What's gone wrong is absolutely essential for us to avoid kind of those pitfalls.

But even if Mike's company Kairos does succeed in eventually building cheaper modular nuclear reactors, nuclear power still has its critics - like, is nuclear power safe?

Evacuations after nuclear meltdowns like Fukushima and Chernobyl were hugely disruptive.

On the other hand, the worst commercial nuclear power plant accident in American history, which is Three Mile Island in Pennsylvania, exposed nearby residents to less radiation than a chest X-ray.

All told, the number of deaths caused each year by nuclear power are incredibly small. They are comparable to that of solar and wind.

On the other end of the spectrum, you've got millions of deaths caused by fossil fuels each year, mostly from air pollution.

Mike says that Kairos is using one of the safer forms of fuels and coolants and that he expects the risk of meltdowns would actually be lower with small modular reactors because there's more standardization.

Plus, you know, he doesn't want a meltdown, either. We have to have a reliable plant in order to have an economically viable product.

And for us, the requirements of protecting the operators and protecting the investment actually fully, like, fully covers everything that's necessary to ensure health and safety of the public.

On the nuclear waste question, that stays highly radioactive for hundreds of thousands of years. Kairos would send the waste to the Department of Energy.

According to the Department of Energy, it's safely storing nuclear waste at more than 70 sites across the U.S. using steel and concrete.

But it's struggling to find places for long-term storage in the face of local opposition.

Another criticism is that given the need to decarbonize quickly, we just don't have the time to wait for small modular reactors to get developed, to get approved by the authorities and start producing energy on a commercial basis.

This could be a decades-long process. We brought this to Mike Laufer, whose deal with Google is that they'll start producing energy in 2030.

The 2030 deadline sounds soon to me, especially given small modular reactors aren't exactly a mainstream technology in the U.S. right now, not to mention that it needs to go through the Nuclear Regulatory Commission.

Why are you confident that 2030 is a doable timeline? So 2030 is certainly an aggressive timeline, and Kairos and Google both share a strong sense of urgency that we need demonstrations, and we need them quickly.

It's not other technologies which are the enemy or the competition. It's carbon. Carbon is the enemy.

So basically, any strategy or technology which allows us to really decarbonize, like, those are going to be highly valued. Those are going to be good things, in my view, for the system overall.

Now, the small modular reactor experiment is bolstered by subsidies, the Inflation Reduction Act in particular, and the fate of those subsidies remains to be seen under President-elect Trump and the new Congress.

For his part, Trump pointed to small modular reactors as a potential answer to long running cost concerns surrounding the energy source. So we might see more companies trying this nuclear option.