Lithium-ion batteries are just about everywhere—they power almost all smartphones, tablets, and laptops. Yet Elon Musk, CEO of Tesla Motors, says he intends to build a factory in the United States three years from now that will more than double the world’s total lithium-ion battery production. The plan is still in its early stages, but already four states are negotiating with Tesla in the hope of becoming the factory’s home.

People have come to expect bold plans from Musk. In addition to founding Tesla, he started his own rocket company, SpaceX, which now delivers supplies to the International Space Station. But even for him, the “gigafactory,” as he calls it, seems audacious.

First, Tesla sold 23,000 cars last year. The gigafactory, which would start production in 2017, would by 2020 make enough batteries for 500,000 electric cars. (It would produce enough batteries annually to store 35 gigawatt hours of electricity, hence the name). Second, battery companies normally announce factories only after they’re funded and a site is selected. And they typically scale up gradually. Why announce plans to build such an enormous factory —especially when electric car sales so far come nowhere close to justifying it?

The project seems more puzzling in light of the hard times at other electric car battery factories in the United States. In 2009, President Obama announced an ambitious $2.4 billion grant program intended to launch an electric car battery industry in the United States. That effort, so far, has failed—factories were built, but sales have been poor because electric car sales have been slow. All of the battery makers involved have struggled (see “Too Many Battery Factories, Too Few Electric Cars”), and one, A123 Systems, went bankrupt.

Musk is betting that Tesla can generate a much bigger market for electric cars. To keep the factory humming, he will have to sell more than 10 times as many electric vehicles in a year as Nissan managed last year (and Nissan has sold more electric cars than any other automaker). Musk has some reason for confidence—last year Tesla sold as many electric cars as Nissan in the United States, even though Tesla’s Model S costs two to three times as much as Nissan’s electric car, the Leaf. He seems to be betting that a huge factory will significantly reduce the cost of making batteries, which remain the most expensive part of electric cars. In the ideal scenario, that cost reduction would help Tesla produce a mass-market car similar in cost to the Nissan Leaf or Chevrolet Volt but that, crucially, will be able to go more than twice as far on a charge (the car would also be able to accelerate faster than the Leaf).

Yet it’s not clear that a huge factory would deliver the needed cost reductions. According to a presentation to investors, it would lower costs by 30 percent. Tesla has a good track record for reducing battery costs (see “Driving Innovation”), and even incremental improvements at conventional factories could reduce costs by 15 percent by 2020, says Menahem Anderman, president of Advanced Automotive Batteries. But it’s unclear where the remaining 15 percent might come from.

Economies of scale could help lower production costs to some degree, but Tesla says the unusual design of the gigafactory, with batteries built from raw materials rather than assembled, will also help.

Usually, the components of batteries are made in many different places. Electrolytes are often made at a large chemical plant and graphite electrodes at a plant that also makes graphite for tires and other applications. The electrolytes and electrodes are then packaged into cells at a plant dedicated to cell making, and the cells are assembled into complete battery packs—with cooling systems and electronic controls—in yet another factory.

Musk plans to bring almost all of this under one roof. Raw materials, processed into electrodes, electrolytes, containers and other parts, go in one end; complete battery packs come out the other. The factory will also be able to take old batteries apart to recycle the materials, and Musk even plans to use solar and wind to help power the factory.

Brett Smith, codirector for manufacturing, engineering, and technology at the Center for Automotive Research, says having control over every part of the process could indeed help reduce costs.

Tesla would need to bring in a great deal of expertise to make this work. What’s more, there are benefits to making different parts in different places. For example, it can be cheaper to make electrolytes in a large chemical plant that makes other chemicals, too. Panasonic, Tesla’s current battery cell supplier, benefits from the know-how of workers in Japan, many of whom have decades of manufacturing experience.

“Manufacturers have tried both approaches. Either approach can work,” says Jack Hu, a professor of industrial operations and engineering at the University of Michigan.

But Hu says such a plant would need to be flexible. “It is possible to build a gigafactory,” he says. “The key lies in the how. Battery manufacturing is a complex process involving many steps. If these steps are all dependent on each other, then the gigafactory would be a bad idea: difficult to run, a lot of down times; difficult to identify quality problems.”

Beyond the technical challenges, Tesla may struggle to convince partners to go along with the scheme. The factory would cost $5 billion, with $2 billion coming from Tesla. If Tesla can’t sell as many cars as it hopes, there would likely be no alternative market for those batteries, making it a risky investment. (One potential market, using batteries for storing electricity power on the grid, is still in early stages of development.)

Whether the gigafactory is actually built, and whatever the final factory looks like, the way Musk has been promoting it may prove to be a savvy business move. Announcing the factory at an early stage, and with an ambitious size, could be good for negotiations with states, especially given the proposed size of the factory. Some states are even reconsidering laws that restrict how Tesla can sell cars in their state, which could help open new markets for the automaker.

Proposing such a huge undertaking might also make it more likely that Panasonic or some other partner will later go along with a less ambitious plan—say, a factory to supply 100,000 cars. “Panasonic can’t afford to lose the business,” says Anderman.

And yet, as grand as Musk’s plan is, it’s worth noting that 500,000 cars is still a tiny fraction of the worldwide auto industry. GM sold nearly 10 million cars last year. If electric vehicles are ever to make a dent in the world auto market, then gigafactories will need to become a reality.