27 Comments

Summary:

The big news in clean power this week is that Apple will build a 20 MW solar farm and a 5 MW fuel cell farm at its data center in North Carolina. Here’s 5 reasons why I think Apple is embracing clean power right now:

BloomEnergy_dudestouching

Seemingly all of a sudden Apple is charging ahead with one of the most aggressive clean power projects for a data center in the U.S. Apple is planning on building a 20 MW solar farm and a 5 MW fuel cell farm at its massive data center in Maiden, North Carolina. Now the real question is why?

I’ve been thinking about the logistics of combining data centers and clean power over the past few years, and here’s 5 reasons why I think Apple is embracing clean power right now:

1). Solar is cheap as heck: Prices of solar panels and cells have plummeted recently, leading to a drop of around 50 percent in prices over the past year. That’s bad news for the solar manufacturers — and has led to a wave of solar maker bankruptcies — but that’s good news for companies, utilities and home owners that are buying solar panels. It’s one of the best times in history to buy solar panels. Particularly when a company like Apple is buying such a sizable quantity (20 MW), they can get an even better deal. A 20 MW solar project planned to be built in Florida this year cost about $70 million — Apple’s spending about $1 billion on the entire data center.

2). Dirty Internet power is bad PR: Greenpeace was one of the first companies to take a close look at the massive data centers being built by Apple, Facebook and Google in North Carolina and the watchdog pointed out that North Carolina has one of the dirtiest power grids in the U.S. — it’s mostly coal (61 percent) and nuclear (30.8 percent). These Internet companies no doubt were attracted to the region because this dirty power is also pretty cheap, at about 4 to 5 cents a kilowatt hour, according to Greenpeace.

But with a growing amount of attention on how dirty this energy mix is, the Internet firms are under greater pressure to bring in their own clean power. In Greenpeace’s report last year it gave Apple a straight “F” for infrastructure siting of its data center, a “C” for transparency, and a “C” for mitigation strategy. Now Apple, like Google and Facebook, have been pushing a lot harder to get low cost clean power into the data center discussion.

3). Control over energy: When Internet companies take a more active role in building energy generation sources like solar and fuel cell farms, they are not only reducing the carbon footprint of their data centers, they are getting more control over a crucial resource that their data centers need. Data centers are major power hogs. And owning the energy source, helps a company like Apple shield at least part of its data center power from potentially rising energy costs.

4). The 100 MW data center: As the size — and power consumption — of data centers rise to the size of Apple’s in Maiden, which will reportedly have a capacity for 100 MW, rural areas and small towns just might not have enough local power generation to fill the need. I’m not sure if that’s the case for Apple in Maiden, but Apple is Facebook Data Centerplanning on adding a sizable 25 MW of its own clean power, which could be a quarter of its power needs. Maiden might not have had the necessary power resources.

5). Fuel cell makers targeting data centers: Fuel cell makers like Bloom Energy, FuelCell Energy, and ClearEdge Power are targeting data center operators with their on site cleaner power sources. AT&T plans to install 75 Bloom fuel cells at 11 of its offices in California and AT&T said it will use the fuel cell power for data centers as well as administration offices and facilities that house network equipment. The U.S. division of Japanese telecom giant NTT, NTT America, said that it plans to install five fuel cells from Bloom Energy at one of its data-center facilities in San Jose, Calif. ClearEdge Power launched a fuel cell line targeted at data-center operators last year.

  1. You missed reason #6: Al Gore is on the Apple board!

    Share
  2. Reason #4 has to win the “stupid statement” prize for this year. “Maiden might not have had the necessary power resources.” Huh? Can you really, honestly, can say it with a straight face, say that Apple built a I-don’t-know-how-many-billions-of-dollars data center and forgot to check if there was enough juice to run the thing? Come on, now. How ridiculous. Maybe they forgot to put bathrooms in too? Didn’t realize people would need them?

    Share
    1. You’re kidding right? Apple’s been planning this huge data center for years. I could see them doing a deal with Duke and the local utility cooperative several years ago and part of it would be to add more power to the region so as not to strain the grid. It’s obviously not that Apple “forgot” about the power.

      Share
      1. Yeah, that’s right. So why did you make that statement that maybe Maiden doesn’t have enough power? Did you read your own article?

        Share
      2. not to mention that this data centre is built for year to come meaning that it is not using as much power today as it will be in 5 years. they may be only using 50MW. A few years of growth will allow the utility to catch up. (assuming that there is or will be a “shortage”)

        Share
      3. Whats funny about this is that I used to live in Maiden, NC about 5 minutes from the data center that Apple built (I was on the complete opposite side of town; it’s that small). While she cites available power she neglects to mention that Maiden is within 60 miles of 5 coal fired plants, 2 nuclear plants, 1 hydro plant and at least 2 large combustion turbine installations. I am only counting Duke Energy facilities as this is the company from which Apple would purchase their power. I am not counting any Progress Energy plants nor am I counting any merchant plants that are in operation in the area. This list is off the top of my head and is by no means conclusive but just goes to show you she did absolutely no research for this article and simply made up the facts she wanted to present.

        Share
    2. Whats funny about this is that I used to live in Maiden, NC about 5 minutes from the data center that Apple built (I was on the complete opposite side of town; it’s that small). While she cites available power she neglects to mention that Maiden is within 60 miles of 5 coal fired plants, 2 nuclear plants, 1 hydro plant and at least 2 large combustion turbine installations. I am only counting Duke Energy facilities as this is the company from which Apple would purchase their power. I am not counting any Progress Energy plants nor am I counting any merchant plants that are in operation in the area. This list is off the top of my head and is by no means conclusive but just goes to show you she did absolutely no research for this article and simply made up the facts she wanted to present.

      Share
      1. Sorry for the double post. I didn’t think it went through the first time.

        Share
  3. Western NC is becoming quite the hub for data centers.

    “The company joins a growing list of companies bringing data centers to North Carolina, following similar decisions by Facebook, Apple, Google, IBM, Time Warner Cable and American Express. NC has a combination of natural and man-made advantages that draw data centers, including relatively cheap electricity and generous state tax incentives.”

    http://www.newsobserver.com/2012/02/22/1876204/att-latest-to-build-data-center.html

    Share
    1. NC is growing for large enterprise facilities, but the fact is, connectivity in NC is still rather undeveloped. But, like most major Internet hubs today – if you build it they will come.. And by they I mean the major Tier I networks.

      Share
  4. Uh, this article is full of it, which is unusual for GigaOM.

    1) Solar may have dropped in price, but it’s still much more expensive than conventional generation.

    3) Solar and wind give you LESS control over your energy supply, because the wind blows and the sun shines at random. When a big cold front blasts past a wind farm, generation first goes up, then goes to zero (the turbines are shut down during high winds), then goes to a minimal number as the wind dies down.

    4) “25 MW of its own clean power, which could be a quarter of its power needs”. Not at night, they don’t, which totally negates the argument that the power may not be available locally.

    5) Most of those fuel cells burn natural gas and the carbon in them has to go somewhere. Also, most of them won’t scale to 100MW.

    Share
  5. Katie,

    I’m not entirely sure where you could have possibly gotten the idea that PV solar is “cheap as heck”. Being that PV solar panels last a maximum of 30 years (most last at maximum 25 years), and the highest capacity factor you can hope for is 19% (and that’s in the Arizona desert, a far cry from North Carolina), PV installations never break even, let alone save, and are entirely a money sink hole.

    Let’s assume Apple spent the same as the installation in Florida you cite: $70 million.

    20MW*24hrs.*365days*(0.19 CapacityFactor)=33.288 million kW-hrs per year is the maximum energy that this installation generates.

    Ok?
    So, we take that and multiply it by 5 cents per kW-hr (as you cited). The total dollars worth of electricity production per year is,

    33,288,000*0.05 = $1.664 million per year.

    If we assume $70 million in upfront costs to build the PV installation as stated above, how long will it take Apple to break even?

    $70,000,000/$1,664,400 = 42.05 YEARS it will take to break even when compared to buying electricity from the grid!!

    Considering that PV panels have a MAXIMUM life span of 30 years, this installation will never even come close to breaking even. In fact Apple is figuratively flushing around $20 million down the toilet.

    All this is assuming ZERO maintenance costs, and a capacity factor that is extremely difficult to achieve even in the Arizona desert. These calculations are laughably conservative. The real total loss of Apple’s cash on this investment is likely to be double that of which I calculated.

    I mean seriously Katie.
    I surprised to see this level of error on GigaOm. Usually, the writers here are near, if not the best, at doing their homework.

    Share
    1. John,
      Your corrections of Katie’s calculations are not correct.
      To get kWh from a system’s kW rating, don’t look at Arizona or theoretical physics, just use the NREL insolation figures for the area you’re interested in.
      For this part of NC, an array will get the equivalent 4.9 hours of full sun per day, year round. Minus losses, Apple’s 20MW system will generate an average of 98MWh each year, not 33MWh. The simple ROI is now 16.8, not 42 years. Once you factor-in incentives and accelerated depreciation, it’s not uncommon to see the ROI drop to about 5 years.
      Also, this ROI will not change regardless of where the cost of electricity goes.

      Share
      1. TO be precise: The above should read “Minus losses, the simple ROI is now 16.8, not 42 years.” I applied losses after the 98MWh. Exact same outcome.

        Share
      2. Do you care to share your calculations?
        The following are based on the numbers you provided.

        98MWh = 98,000kWh
        98,000kWh*$0.05=$4,900 dollars per year of electricity.
        You are very confused

        Share
      3. However, if you meant 98,000MWh per year, you are even more confused, as that would be a capacity factor of 56% which is beyond laughable.
        Did I mention that I’m an electrical engineer.

        Share
      4. Also, just FYI a 56% capacity factor with regard to Solar PVs, literally means that you will get full direct sunlight 56% percent of each 24hr period, which is absurd.

        To be precise:

        4.9hrs/24hrs =~20% or in other words, very close to the 19% that my calculation was based on. This 20% also assumes that the panels suffer no degradation over time. (which is severe, and usually even significant out-of-the-box)

        Share
      5. Aha! I finally see your mistake.

        You made the following calculations:

        4.9hrs per day sunlight * 20MW = 98MW-hrs except that figure is per day not per year. To find output per year we can take

        98MW-hrs * 365 = 35,770 MW-hrs per year! (roughly equivalent to my calculation)

        (In other words, even assuming IDEAL panels, and conditions, and ZERO maintenance costs, the ROI (or break even point) is north of 40 YEARS!!

        Sorry, Louis. I can’t believe it took me so long to figure out what you did wrong.

        Phew…we can all go home now…show’s over

        Share
    2. Your ROI assumes that NC electric rates stay at $.05/kwh for 40 years.

      I suggest you recalculate w/ rates moving to *whatever you suggest in 2052*.

      Share
      1. There is no need. My calculations do not take into account any maintenance costs, or panel efficiency degradation over time, which is significant. Also, a capacity factor of 19% is unrealistic to say the least. 12% is still on the high side but more realistic. At 12% and factoring in all other costs, and degradation, the ROI gets to well over 100 years. Remember, if the ROI is even a hint over 30 years (because the panels will have to be completely replaced at this point) the true break even point goes to infinity.

        Combining all these factors will more that make up for any reasonable jump in electricity costs.

        Share
    3. Actually, given that they probably got or will soon get a cash grant worth 30% of the cost of the project, say $20 million, as well as various tax breaks and considering that Apple is likely billed for time-of-use energy, ie what they buy at 2pm costs a lot more than at night, this may be a financially prudent step.

      Share
      1. I believe the example in Florida of $70 million was after all breaks and incentives

        Share
  6. Don’t forget that solar PV loses approximately 2% of it power producing capacity per year through degradation of the panels. If starting at 19%, after 25 years its power producing capacity is reduced to 11.5%.

    Share
    1. Thank you, yet another thing that my calculation didn’t take into account.

      Share
  7. Ronald Vokoun, DBIA, LEED AP BD+C Thursday, February 23, 2012

    Katie,

    I think you hit on several good points. Some might not have a great ROI right now, but that can change quickly and Apple is looking long term. The thing that can’t be underestimated is the power of branding. Does anyone really want Greenpeace riding them for not applying sustainable practices to their data center? Ignoring sustainability will be at your business’ peril!

    Share
  8. ROI needs to include estimated increase in local electric rates over 20 years. (I’m assuming the solar install will be upgraded w/in 20 years.)

    If we spend $70M today, what’s the cost of electric it’s replacing in 2017, 2022, 27, 32?

    That’s the true savings in electric cost that should be plugged into ROI thinking.

    Share
    1. Assuming full replacement w/ in 20 years. The calculation would be as follows: (assuming ideal panels, ideal conditions,a 12% capacity factor, and zero maintenance costs)

      $70,000,000 / 20 years = $3.5 million per year

      20,000kW * 24 * 365 * 0.15 = 21,024,000 kW-hrs per year

      $3.5 million / 21,024,000 kW-hrs =~ $0.17 per kW-hr or 3.5 times the current cost of electricity. And remember this does not take into account any maintenance cost at all. This assumes you build the thing and let it sit for 20 years. Truth is solar PVs require a lot of maintenance, tweaking, and management.

      Share

Comments have been disabled for this post