#1 cleantech news, reviews, & analysis site in the world. Subscribe today. The future is now.


Published on February 5th, 2016 | by Joshua S Hill


Data Centers Using 30% More Coal As Microsoft Takes Things Underwater

February 5th, 2016 by  

Despite best intentions to go green, data centers have underestimated their coal usage by 30% or more, according to a new report.

Data-services companies like Apple, Google, Facebook, and Amazon are well known for their recent efforts to transition their entire operations to 100% renewable energy. Google especially has been making a strong push towards such an idealistic goal, just this past December purchasing 842 MW worth of renewable energy through Power Purchase Agreements to power their data centers.

Intelligence company Lux Research suggests that data centers around the world are using more than 90 billion kWh of electricity annually — enough to power New York City twice over.

Under-representing Coal

However, the problem, as Lux Research finds it, is that data center companies are using outdated and obsolete data tools for calculating emissions in their purchases from the power grid, and are therefore underestimating the amount of electricity they are using that is sourced from coal.

Lux Research created a new analytical tool that is more up to date and precise than the outmoded method currently used by operators, who have previously had to rely on the US Environmental Protection Agency (EPA) Emissions & Generation Resource Integrated Database (eGRID) to estimate their emissions.

“Our team of data scientists analysed the North American electric grid, improving the accuracy of carbon reporting by a factor of 80,” said Ory Zik, Lux Research Vice President of Analytics and the team leader of Lux’s energy benchmarking. “The results show that many sites are far more reliant on coal than reported – notably, they include many large data centers.”

“For example, we found that Google underestimates its dependence on coal in four out of seven data centers, in particular at its Berkeley County, S.C. location.”

Google Data Centers

Lux Research-2

When comparing eGRID with the new Lux Grid Network Analysis (GNA) tool — which divides the US grid into 134 regions, instead of the previous 24, while making use of data provided by the US Energy Information Administration, which is updated monthly, as opposed to every three years — both Google and Amazon are underestimating the use of coal. Specifically, four of Google’s seven US data centers use more coal than reported by eGRID, subsequently increasing the company’s emissions levels. Amazon, though less transparent about how it calculates its emissions, uses approximately 43% electricity from coal, not the 35% that is inferred from using eGRID.

Amazon Data Centers

Lux Research-3

As Lux Research concludes, “The changing grid drives the need for better tools,” which means that investments into new electricity generation will need to be met by new advances in data collection and analysis.

Microsoft Takes the Cloud Underwater

Microsoft-2While data center operators are going to have to get their heads around the possibility their operations weren’t as clean as they had envisioned, Microsoft researchers have decided to further complicate the imagery of “the cloud” by dropping a data center underwater.

Project Natick was a research project undertaken by a group within Microsoft Research that focuses on special projects. “We take a big whack at big problems, on a short-term basis,” explained Ben Cutler, Project Natick’s project manager. “We’re a small group, and we look at moonshot projects.”

Project Natick’s goal was to put a data center underwater that wouldn’t require maintenance by people — something the everyday data center is reliant upon. In the end, the researchers have learned a lot about how to make data centers more sustainable, while at the same time discovering ways moving data centers underwater can possibly speed data transmission and cloud deployment. Microsoft even believes that “someday, datacenters could become commonplace in seas around the world.”

The idea spun out of a paper that was presented at ThinkWeek. Sean James, one of the paper’s authors, had served in the Navy for three years on submarines. “I had no idea how receptive people would be to the idea,” he said. “It’s blown me away.”


“What helped me bridge the gap between datacenters and underwater is that I’d seen how you can put sophisticated electronics under water, and keep it shielded from salt water,” James continued. “It goes through a very rigorous testing and design process. So I knew there was a way to do that.”

“This is speculative technology, in the sense that if it turns out to be a good idea, it will instantly change the economics of this business,” said Norm Whitaker, who heads special projects for Microsoft Research NExT. “There are lots of moving parts, lots of planning that goes into this. This is more a tool that we can make available to datacenter partners. In a difficult situation, they could turn to this and use it.”

Tags: , , , , , ,

About the Author

I'm a Christian, a nerd, a geek, and I believe that we're pretty quickly directing planet-Earth into hell in a handbasket! I also write for Fantasy Book Review (.co.uk), and can be found writing articles for a variety of other sites. Check me out at about.me for more.

  • Bob_Wallace

    I picked Saudi Arabia simply because it was the first country that came to mind when I thought about who had no hydro resources. If Saudi Arabia bothers you then read in Belgium.

    “you just have to admit it, wind and hydro are better solutions than solar PV for addressing climate warming..”

    No. What one has to realize it the fact that renewable resources are not evenly distributed across all regions and all seasons.

    Solar works well in Massachusetts in the summer. If there’s a summer wind drop off then a mix of wind and solar is going to work better than wind alone.

    On the linked graph hover over Arizona and over New Mexico. Look how the ‘best mix’ changes between wind and solar for two adjacent states.


    Look at Mass. 22.3% solar. 13% onshore wind. 1.4% hydro.

    • thinkmorebelieveless

      Back to my original question about the sunlight-to-heat effect of solar PV. Please answer this: at current consumption rates, if the only source for electrical generation on the planet was solar PV would earth’s atmosphere remain unchanged (as compared to no solar PV at all) or would we be seeing warming ? I believe that if wind or hydro were the only electrical generators the atmosphere would remain unchanged.

      • Bob_Wallace

        Let’s take this back to a pure state. There is no electricity and we invent it for the first time. If we invented only hydro the Earth might cool a tiny bit due to more sunlight being reflected off the bodies of water we would create. Only wind might cause about no change. Only solar might cause a tiny bit of warming.

        Tiny bit because some of the sunlight hitting the panels would not be reflected back out as light but be converted into heat and a small amount of that heat would be trapped inside the Earth’s atmosphere.

        How big is “tiny”? In the map below the small green rectangle is the amount of the Earth’s surface that would be covered with solar panels needed to supply all the Earth’s energy needs.

        Now, are you trying to figure out if using solar will somehow contribute to global warming? Divide those rectangles into half and shrink them some more. We might get 40% of our electricity from solar. Perhaps closer to 30%.

        Nothing to worry about. We worry about the greenhouse gases that are trapping the heat and we use the tools we have to let us quit fossil fuels.

        • thinkmorebelieveless

          Would it be correct to say that the magnitude of solar PV’s sunlight-to-heat effect is equivalent to its total electrical output converted into heat ?
          Changing from the pure state that you describe above to our present world condition of persistent legacy GHG’s, even if we stopped FF use immediately and went to all solar PV would we still have a warming condition ?

          • Bob_Wallace

            ​Given that the Earth was in a long term gradual cooling phase prior to humans botching things up there might be enough extra heat produced to offset the beginning of the next ice age by 0.374 seconds.

            Obviously a guess….​

          • thinkmorebelieveless

            could you answer my question about solar PV’s sunlight-to-heat effect being equal to the heat value of the electricity produced ?

          • Bob_Wallace

            ​How are you using the word “value”?​

          • thinkmorebelieveless

            electrical kW’s being converted into BTU’s ( which heats the air that warms the atmosphere)

          • Bob_Wallace

            OK, replacing “value” in your question with your definition in the most recent comment I get –

            ‘could you answer my question about solar PV’s sunlight-to-heat effect being equal to the electrical kW’s being converted into BTU’s’

            You’re asking if more long wave energy is turned into short wave energy at the panel level than is released once the energy contained in the electricity produced by the panels is used and converted into short wave energy?

          • thinkmorebelieveless

            The question is: does all the extra solar energy that gets converted into electricity by solar PV ( extra meaning in excess of the “natural total” ) ultimately end up heating air ?

          • Bob_Wallace

            The electricity generated by solar panels, wind turbines and hydro plants ends up as heat.

          • nakedChimp

            If there would be no PV panel being hit by a solar ray it would hit earth.
            Earth (in solid places) has an average albedo of 0.1-0.4.. so take the middle – 0.25 (*). This means 25% of the incoming radiation is reflected, the remainder absorbed and then released as long wave radiation.. heat.
            PV panels have an albedo of about 0.3 (**).. this means they reflect 30% of the incoming radiation straight back, no heat. If PV panels cover something with an higher albedo, they increase the heat in that spot and if they cover something with a lower albedo (bitumen, tar-roof-covers, etc) they will decrease the heat conversion in that spot.
            Of the remaining 70% of radiation that has been absorbed, 15-20% are converted into electricity to replace FF energy sources, but finally all of that 70% is converted into heat (except maybe radio-waves or night lights from cities leaving the planet).

            The answer to your question is: It depends. IMHO I’d say it’s a wash or at least negligible.

            *) https://en.wikipedia.org/wiki/Albedo#Terrestrial_albedo

            **) no straight answer, I can quote this:
            or something like this:

          • thinkmorebelieveless

            Thanks for the links. It seems to me that wind and hydro cannot add to atmospheric heating since they do not change the amount of solar energy absorbed. The sun does its normal heating of the wind and water then wind generators and hydro harvest some of this energy and move it to electrical devices which make the heat……no net heat gain. Solar though harvests energy additional to the natural albedo amount, generates electricity from the additional solar energy which then becomes additional heat. I am thinking that all the electricity made from solar PV becomes this additional heat…therefore a net heat gain.

            I will accept that where the solar panels are installed there will be no heat gain from albedo affects. Consider though that if the electricity generated by the solar panels was powering heaters sitting next to the panels wouldn’t there be a heat gain compared to not having the panels installed at all ?

          • nakedChimp

            If the albedo of the panels is less than what they cover, then less radiation is sent back into the atmosphere.

            The absorbed energy (1 – albedo) is converted into electricity (depending on panel efficiency) and the remainder into heat, right there.

            Depending on what you do with the absorbed energy (heater vs laser beam into space) you get more or less heat from the absorbed part, at least 56% in any case though (1 – albedo) x (1 – PV efficiency).

            The panels are always just worse than the difference between albedo before and albedo with panels though..

            So if you cover an area with albedo of up to 0.3 it’s cooling, if you cover areas with albedos of up to 0.56 it depends what you do with the electricity and if you cover areas with albedos above 0.56 it’s definitely heating.

            As for the effect itself you might want to check out some data in regards to total energy balance:
            http://hyperphysics.phy-astr.gsu.edu/hbase/thermo/radfor.html#c1 (the graph in this one shows a net heating effect for solar energy, but there are some cooling human activities that would need offsetting it seems – land use, aerosol effects 😉

          • thinkmorebelieveless

            Thank You

  • Steven F

    With renewables growing at a rapid rate and coal power shutdowns increasing, both models are likely overestimating current coal CO2 emmisisions.

  • ROBwithaB

    My personal planet-saving solution to this particular problem is simply to avoid going onto facebook.

    • nakedChimp

      wasn’t porn the biggest bandwidth hog?

      Besides, who says they can’t build the datacenters more northwards.. cool climates will love the waste-heat and the fast internet 😉

  • Dragon

    I don’t really understand why the underwater data centers angle was thrown in to this article. I also don’t understand the advantage of putting them underwater. Article mentioned faster data speeds… but why? Does it save energy? I could see the water maybe acting as a better heat sink to reduce energy use for cooling… But any time you throw something in salt water it needs much more maintenance than keeping it on land, so it would take a big savings in energy to justify the cost of maintenance.

    • MP

      As most of the world population live by the coast having a shorter distance to travel the data would get to people faster. yes, they randomly cut part of the underwater article here. But, in fuller articles it said they didn’t have to raise the capsule for 5 years before maintaining it and maybe putting the servers in a new inclosing capsule. Also it would save tons as the main power use in a data center in cooling the servers.

    • Steven F

      A lot of energy is used to cool data centers. After all if the computers get hot they will fail. In the ocean if you go deap enough you will always find water at the slightly above temperature of water. So putting the data center underwater could eliminate the need for cooling.

  • While 90 billion kWh sounds like a lot, it is about 0.4% of global electricity consumption.

    • Jenny Sommer

      That is a lot.
      Data centers would rank 34th place under countries.

  • MP

    why not use the excess heat from data centers to preheat the water for power plants?

  • Marion Meads

    We are already heating up the world’s oceans which has been buffering the dramatic effects of CO2, while acidifying it too. Ocean temperatures affect our climate much more so than air temperatures.

    Now we want to heat it up some more and directly by treating it like an infinite heat sink to cool off the overheating data centers? Use the heat or radiate it back to outer space geniuses!

    • JamesWimberley

      The heat gets into the environment anyway. The atmosphere and the sea are in heat equilibrium so I can’t see that it makes much difference. If the electricity is extracted from sunlight or wind, it’s all just recycling within the natural total. There is little heat reaching the earth’s surface from the Earth’s core; the heat is basically from the sun.

      • thinkmorebelieveless

        Doesn’t solar PV upset the “natural total” by not only turning sunlight into heat at the solar panel (same as heating the bare ground) but it is also extracting additional sun energy to generate electricity. This electricity ultimately turns into heat. Therefore doesn’t solar PV heat the earth more than if solar PV wasn’t there ? I would think that wind and hydro do not have this effect since they do not increase the sunlight-to-heat effect, they just move natural amounts of sun energy around with their electrical generation.

        • Bob_Wallace

          A little less energy is reflected back out but that’s far more than offset by the reduction in GHGs.

          • thinkmorebelieveless

            Yes I agree that solar PV is far better than GHG emitting generators. But within the renewable options of solar PV, wind, and hydro aren’t wind and hydro better than solar PV because they do not have this sunlight-to-heat effect ? If so, should we be maximizing our wind and hydro resources and deploy solar PV where wind and hydro resources do not exist ?

          • Bob_Wallace

            Solar produces during peak demand hours. I produces when wind is generally blowing less. That means we need less storage.

          • thinkmorebelieveless

            I am sitting here in Massachusetts in February and solar PV is not producing at peak demand hours (even after I remove the snow from the panels. Greatest demand is in the dark evening and night for lighting, heat pumps, etc. All the world is not the sunbelt !

          • Bob_Wallace

            Congratulations, Captain Obvious.

          • thinkmorebelieveless

            There are millions of us Captain Obvious’ that have to provide electricity while dealing with winter. Why do you make statements like “solar produces during peak demand” when you know it is not true in many places ? Bob, you just have to admit it, wind and hydro are better solutions than solar PV for addressing climate warming……..and all three of them as a triad are magnitudes better than FF.

          • Bob_Wallace

            ​News Flash!!!

            Hydro doesn’t work in Saudi Arabia.​

            (Do I need to explain to you what that means?)

    • Nolan Thiessen

      90B kWh is less than 1 millionth the amount of energy the oceans have gained in the last decade.

    • ROBwithaB

      Of course, it would still be far more efficient to cool the servers directly by using water as a heat sink, than to burn a bunch of coal somewhere to power a turbine (at something less than Carnot efficiency) convert the motion to electricity, transmit the electricity a few hundred kilometres, use that to run a motor that powers a compressor to push some refrigerant gasses through a bunch of tubes.
      Even ignoring the CO2 produced, the coal-fired cooling is dumping a LOT more heat into the atmosphere.

      • nakedChimp

        Neat list.

Back to Top ↑