Can Bloom Boxes Power Data Centers?

9 comments

A look at the Bloom Energy Servers installed on the eBay campus in San Jose.

Is the Bloom Energy Server the future of data center power? Or is it the latest promising energy technology to fall short of the economics and scale required to support major data centers?

Last week Bloom Energy officially unveiled its fuel cell, also known simply as the “Bloom Box.” The company has deployed its units at a lineup of blue-chip customers including Bank of America, Coca-Cola, eBay, FedEx, Staples and Wal-Mart.

Early adopters are enthusiastic, including major players in the data center space. “Distributed power is a big deal,” said Google co-founder Larry Page. “I’m a big supporter of this. I’d love to see us having an entire data center running on this some day.”

Can a Bloom Installation Scale?
Google’s data centers are believed to use large amounts of power - 20 megawatts, and perhaps much more. That kind of scale has been a major barrier for companies seeking alternatives to utility power. Each Bloom Energy Server provides 100 kilowatts of power in roughly the footprint of a parking space.

The Bloom Box contains an array of fuel cells that can converts air and nearly any fuel source – ranging from natural gas to a wide range of biogases – into electricity via a clean electrochemical process, rather than dirty combustion. Even running on a fossil fuel, the systems are approximately 67 percent cleaner than a typical coal-fired power plant, according to Bloom.

But what about the economics? At the current output of 100 kilowatts, it would require 100 Bloom units to support a 10 megawatt data center. At a price of $700,000 to $800,000 per unit, that’s at least $70 million.

Bloom’s customers say the boxes’ cost works out to about 8 to 10 cents per kilowatt hour. But that’s only when state and federal tax incentives are included, as noted by Lux Research.

Subsidies Make the Math Work
“Bloom’s California customers achieve the quoted electricity costs only because they pay for just half of the system’s capital expense, based on the generous 30% U.S. federal tax credit and the $2,500/kW California rebate,” Lux notes in its analysis. “Without incentives, we calculate electricity would cost $0.13/kWh to $0.14/kWh, with about $0.09/kWh from system cost and about $0.05/kWh coming from fuel cost. Note that this is high compared to average retail U.S. electricity costs of roughly $0.11/kWh.”

Amid all of the media coverage of the Bloom unstealthing event, some of the most detailed information came from this one-on-one interview Michael Kanellos of GreenTechMedia conducted with Bloom co-founder and CEO K.R. Sridhar. The interview runs about 8 minutes.

About the Author

Rich Miller is the founder and editor at large of Data Center Knowledge, and has been reporting on the data center sector since 2000. He has tracked the growing impact of high-density computing on the power and cooling of data centers, and the resulting push for improved energy efficiency in these facilities.

Add Your Comments

  • (will not be published)

9 Comments

  1. Rex

    Correction: "At the current output of 100 kilowatts, it would require 100 Bloom units to support a 10 kilowatt data center." You probably mean "... support a 10 megawatt data center."

  2. Paul Tabo

    Very impressive. A scale down in KW hours, economy in production, increase in efficiency makes the system a 21st century device.

  3. Rex

    Bloom Boxes probably won't revolutionize Data Center power, but they could expand options for Data Center operations and locations. Large Data Centers are often built where electricity is cheap, like the Pacific Northwest USA. Could Bloom Boxes enable data center construction where natural gas is cheap, like Texas or central Canada? You could use Bloom Boxes for primary power, and grid power as backup power. Bloom Boxes are small enough to make N+1 and other redundancy schemes feasible. And you avoid the capex and opex for traditional backup generators. Bloom Boxes, like most fuel cells, probably have a long startup time, making them poor choices as backup generators. Bloom Boxes could enable smaller data center construction feasible where grid electric power is limited (e.g. can't get 200 MW in an existing building or saturated downtown location), but natural gas is readily available.

  4. I agree with Rex. Large-scale data centers are typically built near areas of low-cost power. In this situation, the Bloom Box might not make sense. I can think of two situations where a fuel cell operation could make sense. First is a situation where the customer has a high cost of power per kw/hr and they have run up against a constraint from a utility and can’t get additional more power into their facility. In this instance, they could feed a sector of their data center with up to a MW of power from some Bloom Boxes and increase their data center density. The second situation is based on a location where power costs are very high. In this instance, they could use the Bloom Box as the primary power feed and the utility as a second source/generator-type feed. If the Bloom Box drops its feed, an ATS could switch on utility to the load. This is basically the peak shifting model for generator use, but instead of just using at the peaks, it would need to be used all the time to make financial sense.

  5. Bob L

    Well first fuel cells can power a data center. FNBO in Omaha Nebraska has run their main bank DC since 1999 on four 200 KVA fuel cells via natural gas. The fuel cells are primary power, flywheels-generators are secondary with utility power being the final source of power. So nothing new there. The Bloom Box is just another fuel cell. Nothing really ground breaking. If the true cost can get to 5 cents a KWH, then it will be worth while. To me this is all hype.

  6. Hgus

    Natural gas is just one of the fuels that can power a bloom box. why not build data centers on or near land fill sites and leverage methane gas. Landfills are located within reasonable driving distance of major population centers, the land would be cheap (nothing else to do wtih it besides another golf course or a park, roads already exist.... Cheap land, cheap carbon nuetral energy. Not a bad pitch for a data center. Seems like a no brainer to me.

  7. Aaron

    .....or build next to a sewage treatment facility.

  8. Mark155

    The article is incorrect the unit needs oxygen not normal AIR to operate. You must first factor in the cost and energy usage to produce Oxygen needed for power generation. This was not included in the article. Once you factor the cost of the fuel cell the cost of the fuel and the oxygen this unit dose not save any energy or on going costs. The article also did not talk about maintenance or replacement costs or MTBF. There are other hidden cost as well both the oxygen and fuel must be filtered as to not contaminate the fuel cell with impurities. Normal LP gas or natural gas would quickly clog a fuel cell if it were nor filtered. To do this you would need filters that could remove the fine partials from the Gases before they entered the fuel cell. These cost also need to be factored as well as down time for matainance. This is a great PR stunt and that is basically what it is. Please don’t ever fall for another Segway type media blitz.

  9. Please. Stop the agony!!!! Has anyone looked up the history of Ballard Power lately? They spent billions trying to commercialize fuel cells. After 30 years of creating precedent setting clean power projects and 12 years involved with hydrogen and every energy production technology dreamed up, I can't believe the naive silliness I am seeing from Silicon Valley types with shameless promotion and egos to big to realize how little they know about creating clean reliable electricity at competitive prices. Affiliate of our team have been doing data center power at highest possible efficiency since 1985. The Bloom Box is a monument to naive arrogance. It is a device that converts natural gas to electricity through a process that strips the carbon monoxide out of methane and uses the net hydrogen to run through a fuel cell. It is less efficient and less clean (overall) than a good internal combustion engine running on anything from ammonia to diesel fuel. We can put together an uber relaible failsafe clean energy system for this application for a fraction of the cost of a bunch of chrome plated gas t5o electricity converters. Compare this to watches. You can get a reliable time piece in the price range of $6.00 or pay up to $6,000,000.00. They both tell time with the same accuracy. IMHO the Bloom Box is a Patek Phillipe gas generator that is quiet and uses a fair bit of gas, with the inherent Carbon Monoxide emissions of a Honda genset that you can buy for less than a dollar a watt of production. Please I invite you to challenge me in a contest of real documentable facts. This might be a good opportunity for the face book guys to prove they are actually a lot smarter than Google, Ebay etc. Make us all prove it. William Ross Williams