BELLEVUE, Wash. » In a drab one-story building here, set between an indoor tennis club and a home appliance showroom, dozens of engineers, physicists and nuclear experts are chasing a radical dream of Bill Gates: a new kind of nuclear reactor that would be fueled by today’s nuclear waste, supply all the electricity in the United States for the next 800 years and, possibly, cut the risk of nuclear weapons proliferation around the world.

The people developing the reactor work for a startup, TerraPower, led by Gates and a fellow Microsoft billionaire, Nathan Myhrvold. So far, it has raised tens of millions of dollars for the project, but building a prototype reactor could cost $5 billion — a reason Gates is looking for a home for the demonstration plant in rich and energy-hungry China.

(Gates, of course, has plenty of money of his own. This year Forbes listed him as the world’s second-richest person, with a net worth of $67 billion.)

"The hope is that we’ll find a country, with China being the most likely, that would be able to build the demo plant," Gates said last year in a conversation with the energy expert Daniel Yergin. "If that happens, then the economics of this are quite a bit better than the plants we have today."

Perhaps one of the most intriguing arguments supporters make about Gates’ reactor is that it could eliminate several routes to weapons proliferation. Iran, for example, says its nuclear program is for peaceful purposes, but it is enriching far more uranium than it needs for power generation. The United States has long said that Iran is enriching uranium to develop a nuclear bomb.

Today’s nuclear reactors run on concentrations of 3 to 5 percent uranium 235, an enriched fuel that leaves behind a pure, mostly natural waste, uranium 238. (A uranium bomb runs on more than 90 percent uranium 235.) In today’s reactors, some uranium 238 is converted to plutonium that is used as a small, supplemental fuel, but most of the plutonium is left behind as waste.

In contrast, the TerraPower reactor makes more plutonium from the uranium 238 for use as fuel, and so would run almost entirely on uranium 238. It would need only a small amount of uranium 235, which would function like lighter fluid getting a charcoal barbecue started.

The result, TerraPower’s supporters hope, is that countries would not need to enrich uranium in the quantities they do now, undercutting arguments that they have to have vast stores on hand for a civilian program. TerraPower’s concept would also blunt the logic behind a second route to a bomb: recovering plutonium from spent reactor fuel, which is how most nuclear weapons are built. Since so much uranium 238 is available, there would be no reason to use that plutonium, TerraPower says.

Countries that do not have nuclear weapons will still need lots of electricity, said John Gilleland, chief executive of TerraPower, and "we would like to see them build something that allows us to sleep at night."

No one disputes that this is a very long-term bet. Even optimists say it would take until at least 2030 to commercialize the technology. What the competition would look like then — wind, solar, natural gas or some other technology — is not clear. If the idea can be commercialized, it is not even clear that TerraPower could do it first.

The engineers working for Gates acknowledge the enormous challenges but say they are convinced that he and they are chasing the solution not only to energy and weapons proliferation but also to climate change and poverty.

"If you could pick just one thing to lower the price of – to reduce poverty — by far you would pick energy," Gates said as he introduced the reactor idea in a speech in 2010. "Energy and climate are extremely important to these people, in fact, more important to them than anyone else on the planet," he added, referring to killer floods, droughts and crop failures driven by carbon dioxide given off in energy production. He illustrated his talk with a photo of schoolchildren doing their homework under street lamps.

Doug Adkisson, TerraPower’s senior vice president for operations, said Gates had "a very humanitarian but very cold assessment" about nuclear power and what it could do. What drives him to nuclear power, he said, are the questions "What have you got, and what can you do to raise the living standard of a whole lot of people?"

Despite its difficulties, some outside experts applaud Gates for trying.

"If you’ve got a huge amount of money, for whatever reasons, you are willing to make a long-term bet, which is not typical of what venture capitalists do," said Burton Richter, a Nobel laureate in physics. "It’s hard to get a 20-year thing from the standard venture capital world," he said, adding that financing projects like TerraPower’s was more typical of governments or sovereign wealth funds.

One-hour meetings with Gates about TerraPower sometimes turn into five-hour meetings, associates say.

In Bellevue, TerraPower is a spinoff of Intellectual Ventures, a company co-founded by Myhrvold that focuses on inventing new products and techniques, among them improved seeds for subsistence farmers and methods for keeping vaccines cold for weeks in places where there is no electricity. But its critics call it a patent troll because it buys large portfolios of technology patents and uses them, they say, to sue software designers, smartphone makers and others.

TerraPower employees work in a building that also houses Intellectual Ventures, which includes a chamber for raising mosquitoes, a test kitchen for developing new ways to prepare and preserve food, and hand-built, high-precision instruments for measuring tiny details of prototype nuclear fuel.

Some of its equipment has more than one use: The nuclear effort shares a supercomputer, one of the 500 fastest in the world, with the vaccine and disease vector section, and a tool that cuts steel with a jet of water propelled to three times the speed of sound is used for various programs.

The design is fundamentally different from that of reactors now in commercial use. Enrichment of the uranium 235 used by existing reactors has left behind hundreds of thousands of tons of very pure uranium 238 around the world.

One of the biggest challenges TerraPower faces is that neutrons — the particles released when a uranium atom is split in a reactor — damage a reactor’s metal parts. In today’s reactors, the problem is manageable because the fuel stays in place for no more than six years and can stand the bombardment. But the TerraPower fuel is supposed to stay in place for 30 years.

"The biggest problem is swelling," said Kevan Weaver, a physicist and TerraPower’s director of technology development. "The neutrons knock an atom out of the lattice, and leaves a hole, and then the holes coalesce and form voids, and the part swells."

So TerraPower’s engineers are experimenting with different types of metals, at different temperatures. In December they will put thousands of samples into a Russian reactor that will irradiate them for six years, with neutrons of the same energy that TerraPower’s reactor would have. At the end of this decade, they will see how the metals’ strength was changed, and predict whether the metal will survive for 30 years.

Another problem is that when uranium is split, some of the fragments are gases. This is tolerable in current fuels, but no fuel could hold a 30-year accumulation.

Designing the core of the reactor is an additional problem. TerraPower engineers call it a "traveling wave reactor," because the area in which the uranium 238 has been converted to plutonium and can be fissioned travels through the core like a wave.

But every time the designers change the thickness or type of metal they are using, the flow of neutrons will change, too, and the 30-year life of the core is so long that the inventory of fission products, some of which absorb neutrons, will also change, as some unstable materials give off radiation and transmute themselves into something else.

To allow the neutrons to travel at a speed that is best for converting waste uranium into plutonium fuel, the reactor uses sodium, not water, to moderate the neutrons’ speed and carry off the usable heat. But hot sodium burns on contact with air.

TerraPower is not alone in pursuing a reactor that will turn waste uranium into energy, and if such a concept can be commercialized, Gates might not be the first to do it. General Atomics, which has decades of experience in nuclear power but is probably best known for producing the Predator drone, is pursuing what it calls an "energy multiplier" reactor module on the same general principal. General Atomics, which is based in San Diego, would use helium, not sodium, however, potentially simplifying some problems.

"You just set it up, let it burn, and it goes," said John Parmentola, the company’s senior vice president.

Like TerraPower, General Atomics is courting the Chinese.