The new World Trade Center towers in New York City will be powered in part by fuel cells. Whole Foods runs some of its supermarkets on fuel cells. Walmart, eBay, Google, Staples, Coca-Cola, and many other major corporations have installed them in the last few years. Many of these companies say that they’re not just using fuel cells to reduce energy consumption and pollution, provide reliable backup power, and attract good publicity. They also aim to save money.

If they’re successful, and many of the initial results suggest that they could be, it will be because the cost of fuel cells has dropped significantly over the last few years. They last longer than they used to, and at the same time, local and national governments have provided generous subsidies. “There are an increasing number of cases where the pure economics of adoption make sense,” says Kerry-Ann Adamson, a research director at Pike Research. But fuel cells remain an expensive way to generate electricity, and they don’t yet make economic sense for all businesses.

Fuel cells, which are available in more than half a dozen varieties, produce electricity directly using chemical reactions rather than by burning fuel to spin a generator, the method used by conventional power plants. Like a battery, they use positive and negative electrodes separated by an electrolyte. But while batteries store all the chemicals needed for the necessary chemical reactions in a sealed package, fuel cells use an external fuel source and air from the atmosphere. As a result, they can run continuously as long as there’s a supply of fuel. Different fuel cells use different fuels. Some require pure hydrogen gas. Others can run on natural gas or even diesel. The first fuel cells were made in the 19th century, and they’ve long been used to generate power in remote areas and on the space shuttle. But they’ve been too expensive to replace conventional power plants in most cases.

Powering commercial buildings with fuel cells can cost hundreds of thousands or millions of dollars. They make the most sense for hospitals, banks, supermarkets, and companies with big data centers—businesses that require a reliable source of backup power and are willing to pay a premium. Backup generators are typically far less expensive than fuel cells: a natural-gas generator might cost 50 cents per watt of generating capacity, while fuel cells can cost over $7 per watt, says Sam Jaffe, a research manager at IDC Energy Insights. The problem is that in most places, local regulations limit the use of generators, so most of the time they’re sitting idle. Fuel cells, which are quiet and clean, can run continuously, so companies can recover their investment more quickly. (The best natural-gas generators can also minimize emissions, and if regulations are modified to take this into account, fuel cells may lose their advantage, Jaffe says.) Fuel cells are particularly attractive for new buildings, which can be engineered to take advantage of the waste heat generated by the cells to provide hot water, heat the building, or cool it with the help of special chillers driven by heat or hot water. One supermarket in New York even uses heat from its fuel cells to keep its sidewalks free of ice. A well-designed fuel-cell system can use 90 percent of the energy in the fuel it consumes. By contrast, even the most efficient power plants are less than 60 percent efficient (and some coal plants are less than 40 percent efficient).

Expensive though they are, fuel cells have also begun allowing some companies to save money. In 2007, Whole Foods Market installed a 200-kilowatt phosphoric acid cell in a supermarket in Connecticut, where subsidies fund 25 to 40 percent of the cost of a fuel-cell system. The fuel cell, which runs on natural gas, is made by UTC Power, a subsidiary of United Technologies Corporation. It provides about half of the store’s power, and the market pays 30 percent less for energy than similarly sized stores that use the same electric utility. Whole Foods has since installed two 400-kilowatt fuel cells and plans more, but not everywhere. “Fuel cells only make sense where there are state incentives,” says Kathy Loftus, Whole Foods’ global leader of sustainable engineering and energy management. Rather than buying the fuel cells, Whole Foods reduced its risk with a 10-year lease, which includes a requirement that the technology produce at least a certain amount of power.

The fuel cells are still “not inexpensive,” says Michael Glynn, communication and marketing manager at UTC Power. But if a business needs power and heat night and day, and its buildings are engineered to use the waste heat, “companies can see payback in energy savings within five years and as low as three years,” he says.

One up-and-coming fuel-cell provider for large businesses is Bloom Energy, which makes what’s called a solid oxide fuel cell. The startup came out of stealth mode last year to a great deal of fanfare, with announcements that its customers included Walmart and eBay. It also faced a healthy dose of skepticism. Several other companies have developed technologies similar to the one it uses, and its initial costs ($700,000 to $800,000 for a 100 kilowatt system) seemed in line with what other types of fuel cells cost. What’s more, Bloom’s cells weren’t designed so that their waste heat could be put to use. Although this makes installation simpler, it eliminates a significant source of energy savings. Yet 200 of Bloom’s systems have been sold and installed so far.

Bloom’s success seems to be due in large part to a successful marketing campaign and federal and state subsidies that cut the costs of its systems in half. This month it announced a new financing plan that could help even more. Instead of buying fuel cells, companies can just buy the electricity the cells produce, with rates locked in for 10 years. Bloom claims that for some companies, this can reduce electricity costs by 20 percent.

For all the cost reductions and financing innovations, fuel-cell companies are still leaning heavily on government subsidies, and it’s not clear how long those will last. “The real long-term challenge is to reduce production costs at the same rate as these subsidies are drawn back,” Jaffe says. He estimates that costs have to be cut in half for fuel cells to be economical without subsidies.