The Tesla founder and private space entrepreneur Elon Musk announced yesterday that Solar City, the solar installation company where he is chairman, plans to acquire a startup called Silevo for $200 million (plus up to $150 million more if the company meets certain goals). And with typical bravado, he also said that the company plans to build a huge factory to produce Silevo’s high-efficiency solar panels, a strategy he claims will make solar power “way cheaper” than power from fossil fuels.

Solar City is one of the country’s largest and fastest-growing solar installers, largely as a result of an innovative business model that allows homeowners and businesses to avoid any up-front cost. If its plans pan out, it will also become a major manufacturer of solar panels, with by far the largest factory in the U.S.

The acquisition makes sense given that Silevo’s technology has the potential to reduce the cost of installing solar panels, Solar City’s main business. But the decision to build a huge factory in the U.S. seems daring—especially given the recent failures of other U.S.-based solar manufacturers in the face of competition from Asia. Ultimately, however, Solar City may have little choice—it needs to find ways to reduce costs to keep growing.

Silevo produces solar panels that are roughly 15 to 20 percent more efficient than conventional ones. They incorporate thin films of silicon, which increase efficiency by helping electrons flow more freely out of the material, and they use copper rather than silver electrodes to save costs. Higher efficiency can yield big savings on installation costs, which often exceed the cost of the panels themselves, because fewer panels are needed to generate a given amount of power.

Silevo isn’t the only company to produce high-efficiency solar cells. A version made by Panasonic is just as efficient, and SunPower makes ones that are significantly more so (see “Record-Breaking Solar Cell Points the Way the Cheaper Solar Power”). But Silevo claims it could make its panels as cheaply as conventional ones if it can scale up from its current production capacity of 32 megawatts to the factory Musk has planned, which is expected to produce 1,000 megawatts or more.

The factory plan mirrors an idea Musk introduced at one of his other companies, Tesla Motors, which is building a huge “gigafactory” that he says will reduce the cost of batteries for electric cars. The proposed plant would have more lithium-ion battery capacity than all current factories combined (see “Does Musk’s Gigafactory Make Sense?” and “Tesla Plans to Start Building Its Gigafactory Next Month”).

One key difference, says Travis Bradford, who directs the Energy and Environment Concentration at Columbia University’s School of International and Public Affairs, is that at least three other solar companies—First Solar, Yingli Solar, and Sharp—have already built one-gigawatt factories. Still, the plant would be much larger than any now in the U.S.

By no means is it certain that Silevo will achieve the cost reductions it’s hoping for without sacrificing quality. But subsidies from the state of New York, where Solar City plans to build its factory near Buffalo, should help provide a temporary boost. New York is putting in $225 million in infrastructure for the factory, which might help offset the damage to U.S. solar manufacturers that the U.S. government says have been inflicted by subsidies in places like China. More subsidies are being negotiated.

Besides, Solar City may need to take the risk of investing in a large new manufacturing plant in New York. Existing subsidies that have helped the company grow quickly in recent years may soon disappear. Meanwhile, utilities may start wanting to pay solar-panel owners less for the electricity they produce, because they’re increasingly concerned about the intermittency of solar power and the stress this can put on the power grid. But Solar City believes that if building huge factories to produce advanced solar panels can bring down costs, the market for solar panels could still grow exponentially.