Solar Power and Data Centers

When I started my blog over a year ago, my focus was solely on making data centers energy efficient. Over time, I started covering other areas related to the original theme, including green IT, cloud computing, energy, and smart grid. Moreover, I cover both U.S. and Japanese markets.

I attended an evening event held by the Keizai Society, which is a group that:

provides programs that showcase specialists with expertise on issues critical for the success of companies based in either the U.S. or Japan and promoting entrepreneurs. Its goals are to provide an informal, informative environment built on respect that allows for mutually beneficial communication and networking.

This time they brought three speakers to discuss green technology.

The first speaker, Tony Seba, presented his pitch on solar energy to fill the gap between supply and demand in the year 2050. ::continue:: Unlike my colleagues at AltaTerra, I am not an expert in solar energy, but I found his presentation both informative and entertaining.

His talk was based on his book Solar Trillions: Create Wealth, Grow the Economy, Save the World (forthcoming). He started with four myths about solar power, and his explanations were credible and backed by a good set of data. These are the four myths:

  1. Solar still needs R&D and is not ready for prime time: During 2007, all San Francisco homes together consumed 1.45 TWh, while solar plants in the Mojave Desert alone from 1985 to 2001 cumulatively generated 14+ TWh of power. Note: The T is for tera—trillion (one thousand times bigger than giga).
  2. Solar is a photovoltaic panel on a rooftop: This is fine, but in order to scale, a large city-block (utility scale) size installation is in place and generating a considerable amount of power.
  3. Power by solar is too expensive: His Pacific Gas and Electric (PG&E) bill indicates 11.5 cents/kWh, as opposed to 9–12 cents/kWh by solar power. Although solar is not awfully inexpensive, it is comparable to the current price.
  4. Intermittent and (sunny) daytime only: One of the leading installations in Spain deploys power storage that last 7.5 hours, and in 2010 this will be extended to 15 hours.

Then he did some math for power demands in 2050. Note that 2050 is about 40 years from now because a typical power plant lasts 40 to 50 years. In 2007, the total power generated in the U.S. was about 14 TWh. The projected power demand in 2050 is 28–35 TWh. Assuming we can maintain the 14 TWh as a base, we still need 16 TWh (a total of 30 TWh).

Where will this 16 TWh come from? Clean energy. Seba likes geothermal energy and so do I. But by 2050, the best geothermal could do is 1.5–2 TWh. His conclusion is that solar power will generate the additional 16 TWh. Power generated by solar worldwide could be 120,000 TWh/year. The 30 TWh we need is only 0.025% of that. In addition, less than 1% of the desert can provide all the power necessary for the entire world. He also predicts that, by 2020, power generation by solar will be cheaper than that by coal.

A good-size data center may require 1–5 MW of power. Very large ones may require power in the 50 MW range. I forgot to ask Seba if one solar plant can or will generate that much power. If, however, that is possible, it would satisfy one of the necessary conditions for selecting a location for a data center. Such a solar plant not only satisfies the power requirement but the power is clean. Of course, we need to satisfy one more condition, which is access to the Internet.

Finally, if you cannot build a data center close to solar power plants, you need to transmit the generated power to where you do build it. It is well known that constructing a new transmission grid is time consuming and very costly. Another problem is energy storage technology. These problems are currently under consideration.


Zen KishimotoBy Zen Kishimoto, PhD, Principal Analyst, Green IT, AltaTerra Research

You can also find this blog post (under the title, Keizai Society Presented “Green Technology and Collaborative Business Opportunities” —Part 1) and more on under Zen and the Art of Data Center Greening.

Previous articleUnlocking laser tools’ potential in c-Si cell fabs
Next articleTokuyama To Build 6,000 Ton Polysilicon Plant
AltaTerra Research is a research consultancy specializing in sustainable business and the commercial marketplace for clean technology solutions. Through research reports, executive events and advisory services, AltaTerra helps forward-looking organizations improve resource-efficiency in their operations and capitalize on new 'green' market opportunities. From our headquarters in Silicon Valley, we have been serving a global base of corporate, institutional, and government clients since 2007.

No posts to display