Since there are no currently active contests, we have switched Climate CoLab to read-only mode.
Learn more at
Skip navigation
Share via:


Maximum impact, minimum cost with molten salt reactors, cryptocurrency, carbon income, and electric self-driving taxis



America has a severe political disadvantage when it comes to dealing with climate change: a large political constituency that denies it exists, and a national legislature that's largely captured by industry lobbyists. We also have cheap natural gas, which emits less CO2 than coal but more methane, and outcompetes cleaner energy sources.

But America also has advantages. We have vast farmlands that could sequester CO2 if managed properly. We have Silicon Valley. We're making self-driving cars. We're early adopters of cryptocurrency. We have startup companies attempting to build new forms of cheap, small-scale nuclear energy, and even develop nuclear fusion. We're not great at rational politics, but we're fantastic innovators.

In fact, we're such great innovators that we're automating ourselves out of jobs. But that can help our climate efforts too.

We start by keeping the government out of the way. Cryptocurrency can motivate voluntary carbon offsets. Electric self-driving cars can give us a short path to reducing transportation emissions. New nuclear designs can give us cheap, zero-emission, reliable energy with fast deployment. We need targeted lobbying to make sure these innovations are allowed.

We implement fee-and-dividend state by state, paving the way for a national system.

As more people lose their jobs to automation, demand for a "basic income" scheme is likely to reach a tipping point. When that day comes, we can fund basic income with a carbon fee, returning fees to the general population.

Above all, this plan emphasizes things we can starting doing soon, at minimal cost and with minimal political resistance. The longer we wait, the harder our task. The more expensive our solutions, the slower we'll deploy them. Our resources are not unlimited, and as time passes, damage from climate change will eat more of our resources.

Which proposals are included in your plan and how do they fit together?

We start right away with ClimateCoin, a new cryptocurrency which mints new coins to anyone who pays to offset carbon. It's implemented on a scriptable cryptocurrency platform called Ethereum, and can offset carbon via existing providers of voluntary carbon offsets.

In particular, America has vast farmlands, which if properly managed could sequester a large amount of carbon. One well-known method for reliably sequestering large amounts of carbon is to burn organic materials in low oxygen, producing charcoal. Once ground and worked into soil, it remains stable for centuries, and in many environments improves the fertility of soil. Carbon-Negative Biochar Economies suggests using cryptocurrency to fund biochar projects.

Biochar isn't the only way to sequester carbon in soil. Topsoil-friendly farming practices can do it as well (though less permanently). So can deep-rooted prairie grasses.  advocates these methods and  suggests funding them with cryptocurrency. Soil carbon is directly measurable, making this approach well-suited to offsets.

Power plant migration

A main point of last year's Carbon-Free Fast was that we can't afford to run our fossil plants to end of life. To close them early, the total cost of energy from carbon-free plants has to be less than the ongoing variable cost of the fossil plant.

We can do this by subsidizing carbon-free capital costs with Climatecoin, applying a carbon fee, and using new cheaper carbon-free technologies.

Fee and Dividend

Charge a fee per ton of emitted CO2, with all revenues returned to citizens, as described by

It may be infeasible to enact this at the national level, for now. But we can start at the state level, just as advocates of marijuana legalization and gay marriage started with states to pave the way for national policy. Some states have already explored carbon pricing. 

Longer-term, the U.S. is an early adopter of automation, machine learning, and AI, and will probably be one of the first countries to suffer high technological unemployment. Many advocates want to counter this with "basic income," which gives everyone a fixed minimum income with conditions.

Funding for basic income has to come from somewhere; if we fund it with carbon fees, it's identical to Fee and Dividend. See  

By recasting carbon fees to emphasize the income side, we change the conversation. For some reason, every country thinks it can only afford a carbon fee if every other country also has a fee. But nobody believes that their country can only afford a social safety net if other countries have them as well. We have no global conferences on welfare or retirement systems. Every country just builds its own system.

Basic income plans typically suggest an income in the range of $10,000 per year. If funded entirely with carbon fees, that would equate to a fee of $500 per ton CO2, which would result in a swift transition away from fossil fuels.

That's not likely to happen soon; that's why we start with Climatecoin and state systems. But basic income advocates are likely to increase in number as technological unemployment grows. Climate advocates should join forces with them.


One category of employment that's in danger of technological obsolescence is professional driving, now that self-driving cars and trucks are in testing on public roads. This provides us another opportunity.

The most efficient way to make cars and trucks carbon-neutral is to make them electric, with carbon-neutral power plants. Replacing our entire vehicle fleet is a daunting task. But with self-driving vehicles, we don't have to replace the entire fleet.

Most of our cars are idle about 95% of the time. A consumer relying only on Uber-style car sharing can save most of the capital expense of owning a car, paying only for energy, a small portion of capital, and a modest profit margin for the car service. By doing it this way we can build an order of magnitude fewer cars. The cars can drive themselves to charging stations, minimizing the inconvenience of long charging times. Once the technology is ready, adoption is likely to be quick.

Once consumers are used to not having cars, there may be more demand for even more efficient public transportation. An excellent and very low-cost solution is  


America has abundant natural gas, which is sometimes considered a "bridge fuel," given its lower CO2 emissions. Unfortunately it emits quite a lot of methane. We can reduce that with

America also has significant potential for wind and solar energy. We should certainly take advantage of that. However, while wind and solar are inexpensive given natural gas backup, if we want to run the country on them we'll have to deal with their inherent variability. That means overproduction, storage, smart grids, and long-distance transmission, all of which add major costs.

The other option is nuclear. Unfortunately, capital costs for conventional nuclear are high. Since nuclear fuel is cheap and nuclear plants last a long time, nuclear's cost per kWh is competitive, but that initial cost is a killer. 

Why Abundant Natural Gas Makes Renewables a Trap

Wind and solar produce power about 30% of the time. Unless we overbuild drastically, the rest has to be covered by fossil or nuclear.

A natural gas plant has a fixed cost of about $18/MW, and variable cost of $54/MW. If it only runs 70% of the time, then we have to divide the capital cost by 0.7 to get the effective cost per MW, thus adding $8/MW to the overall cost.

A conventional nuclear plant is $81/MW fixed cost and only $12/MW variable. If it runs only 70% of the time, its effective capital cost increases by $34/MW.

Thus, transitioning to wind/solar makes it harder to add conventional nuclear, and further entrenches natural gas. Lucky for us, we know how to make cheaper nuclear.

Escaping the Trap with Cheap Nuclear Fission

Decades ago, the U.S. invented molten salt reactors, and ran one with great success for four years at Oak Ridge National Laboratory (ORNL). Then the molten salt program lost the funding battle, and was dropped.

Now, at least three startup companies in the U.S. and one in Canada are attempting to revive the idea. There's also a major MSR development program in China. Oak Ridge National Laboratory and the Department of Energy have been assisting these efforts.

Molten salt reactors are advocated in the proposals:



Some molten salt designs are advanced breeders, which we don't really know how to make yet. But two companies are pursuing simple non-breeding designs, which are simple extensions of the reactor we had running decades ago.

Terrestrial Energy

This is a company in Canada, producing the Integral Molten Salt Reactor, or IMSR. The reactor is a permanently sealed vessel, built in a factory and small enough to be transported by rail, ranging in size from 80 to 600 MWt. The reactor has a lifetime of seven years, then is shipped back to a refurbishing facility.

The most troublesome fission products (cesium, iodine, and strontium) are chemically bound in the liquid fuel. Safety is entirely passive based on simple physics. There's nothing to drive a chemical explosion, and the fuel and coolant is at ambient pressure. The reactor is strongly proliferation-resistant and requires only low-enriched uranium.

The reactor uses only 1/6 as much uranium as conventional reactors. Assuming a uranium price ten times higher than now, there are estimated economical reserves sufficient to power civilization for thousands of years. We can expand nuclear power production by a factor of six before needing additional enrichment.

The company estimate capital cost at about $2/watt, with insignificant fuel and operational costs, for a cost per kWh competitive with natural gas. They are privately funded and intend to have their first commercial reactor operating by early next decade.


ThorCon's reactor is similar to the IMSR, at a somewhat larger scale, designed to be built in high volume at low cost by shipyards. A 1GW plant would be delivered in 20 modules by barge. Like the IMSR, the reactor cores are sealed and replaceable, in this case with only a four-year lifetime. The reactor uses no new technology, relying on the research that was already done by ORNL.

A 1GW reactor would be 1/4 the size and less complex than a ship which costs $89 million and takes nine months to build.  ThorCon has calculated that a single large shipyard could easily produce reactor capacity of 100 GWe per year, with an estimated cost of $1/watt and $.03/kWh.

ThorCon is not the only group advocated shipyard construction of nuclear reactors. There's also which advocates more conventional light-water reactors, built in shipyards and installed offshore for easy siting and limitless cooling.


Nuclear regulations in the U.S. are very inflexible, restrictive, and slow to change. They are designed entirely for conventional reactors. We can't develop or deploy cheaper, safer designs unless we reform our regulatory regime. We need regulations that approve a mass-produced design, without requiring separate approvals for each installation, and which are flexible enough to accommodate new reactor types. This problem is addressed by

Long-term Nuclear

These simple reactors are fuel-efficient but since they're not breeders, they couldn't run civilization for more than a few centuries. But we could go much further with more advanced technologies, including fast breeders, thorium molten-salt breeders, and fusion. Proposals include:

The U.S. Integral Fast Reactor project was nearly complete in 1994, when it was shut down by the Clinton administration. As a breeder, this design could run civilization for many thousands or even millions of years.

MIT recently released a design for a more compact tokamak fusion reactor, using stronger magnetic fields, which could be built in under a decade for $5 billion, and produce 250MWe. MIT currently operates a tokamak with strongest magnetic field of any in the world; unfortunately Congress is on the verge of shutting down their fusion program.

There's also the Dynomak program at UW, using a spheromak, which is a somewhat more advanced version of the tokamak. They need about $10 million to test whether their design will scale; if so it's straightforward engineering to production.

Private investors are funding more speculative attempts at achieving cheap practical fusion within a decade, including Tri-Alpha, General Fusion, LPP's focus fusion, Helion, and a project at Lockheed. The MagLIF project at Sandia National Labs also has potential for near-term fusion power. If we do the research, and we're very lucky, we could have a breakthrough in cheap, clean energy. But if not, molten salt reactors can still do the job.

Explanation of the emissions scenario calculated in the Impact tab

We assume that in 2025 we start rolling out ThorCon reactors at $35/MWh, about a third of the existing price for nuclear power. We also assume continuing improvement in renewables, and further cost reduction by 2035.

We assume a $100/ton carbon price starting in 2025. (A smaller initial price rising over time is more likely, but that's not an option in the model.)

EnROADS has a maximum 10% improvement in transportation. With a switch to electric taxis powered by zero-carbon energy, we should do much better than that.

(Note: meant to tweak this model further, but it wasn't working for the last couple days of the contest.)

What are the plan’s key benefits?

Low cost: the reactors are projected to be competitive with natural gas and significantly cheaper than coal.

Low technical risk: the nuclear designs described here don't use new technologies. They're straightforward applications of research done decades ago.

Reasonably low political risk: given our low costs, we can have a large impact without a carbon fee, and can replace most U.S. energy production with only a modest fee. At the same time, our emphasis on carbon income could build support even for a higher fee. We also have a modest  economic incentive that requires no government action at all.

There is significant political resistance to nuclear, but for many people, the reasons for that resistance are safety, weapons proliferation, and waste. These designs have dramatic safety advantages that are easily described to the general public. They are very proliferation resistant, and they pave the way for more advanced designs that can eliminate most of our existing waste.

What are the plan’s costs?

ThorCon estimates capital cost under $500/kW, producing energy at $35/MWh, competitive with natural gas. A single shipyard can produce capacity of 100GWe per year. Total U.S. energy consumption is about 3.5 terawatts, so that one shipyard could replace all existing U.S. energy production in 35 years, at a cost of $50 billion per year.

Note that's capital cost, which will be paid back with electricity sales. Since the cost is competitive with natural gas, a modest carbon price will be sufficient to displace all new fossil plant construction. 75% of U.S. coal plants are within ten years of retirement; to minimize cost we need to act quickly, before replacement fossil plants are built.

Conversion to self-driving electric taxis seems likely to happen on its own, as long as not restricted by regulation. Skytran costs $10M per installed mile, about a tenth of light rail.

What are the key challenges to enacting this plan?

The nuclear technologies advocated here are very conservative, with a very short development path. Our challenges are political.

The NRC is our primary obstacle to advanced nuclear development, and reform may be difficult, though the NRC chair is beginning to assist with reform plans.

State and local jurisdictions may interfere with self-driving cars and "unlicensed" taxis services.

Entrenched fossil interests may be very influential in limiting cabon-free energy and transportation. We'll need clear economic advantages to overcome that.

New regulations in New York and California may slow the adoption of cryptocurrencies.

For the most part, we're asking for only a modest carbon free and the freedom to innovate. That freedom is supposed to be America's strength. That's still true in many industries, less so in others; in nuclear technology China allows far more innovation now. Nevertheless, we have many companies and investors that want to forge ahead; we just need to let them do it.


Climatecoin can launch as soon as Ethereum is sufficiently established so carbon offsetters will take payment in Ethereum's native currency, perhaps in a couple years. Earlier launch would be possible with a transitional company that converts ether to dollars and buys offsets. 

MIT's nuclear regulatory plan suggests a pilot plant in 2020 and full-scale licensing of unconventional nuclear by 2025. Until then, nuclear startups can do R&D in friendlier regimes.

Electric self-driving cars should be in volume production by 2025. 

ThorCon and Terrestrial Energy could have prototype reactors operating by the early 2020s. ThorCon says it could easily roll out 100 GWe capacity per year from a single shipyard, with no restriction due to fuel supply.

Politics makes the carbon fee timeline difficult to predict, but working state-by-state and taking advantage of concern over mass technological unemployment should help.

Related plans


In addition to references in the source proposals...

ThorCon Power and presentation

Terrestrial Energy and presentation

Transatomic Power (another molten-salt company, founded by MIT students)

Overview of molten salt reactor projects

How to Make Nuclear Cheap (pdf)

Recent Harvard presentation on tackling climate change with advanced nuclear

Google engineers on the need for energy breakthroughs

75% of coal plants retiring within 10 years - let's not miss this opportunity

MIT fusion design: arxiv and press release


A good introduction to near-term mass technological unemployment is the 2012 book Race Against the Machine, by MIT professors Erik Brynjolfsson and Andrew McAfee.

One approach to carbon-friendly farming

Tackling climate legislation at the state level

EFF on the trouble with the California Bitcoin license

Google hopes to make self-driving taxis available to the public in four years

The first Skytran demonstration should be complete this year