Earlier this month Stanford University announced that its endowment would divest itself of any investments in companies "whose principal business is coal." It was, they said, a principled move reflecting their concerns about global warming. Stanford's timing was auspiciously contemporaneous with the release of the White House's new 800+ page global warming report, Climate Change Impacts In The United States.
You do not have to have an opinion on global warming, or coal itself, to see a logic problem with the posture of eschewing companies that produce coal, but supporting companies that use lots of it. And we are not referring to coal-burning electricity utilities (though logic would argue Stanford divest there too), but to the legion of tech companies that are in many cases right in Stanford's backyard, and that are some of the world's biggest users of coal-by-wire.
The logic: The principle business of the tech community is anchored in bits. But all bits are electrons (or their quantum cousins, photons) and thus the Internet's monthly exabytes of data traffic consumes vast quantities of electricity. And coal remains the principle source of electricity for the U.S. and the world.
Companies such as Amazon, eBay, Facebook, Google, HP, IBM, Microsoft, Oracle, Rackspace, Salesforce, Twitter, and Yahoo, consume huge amounts of electricity from the grid, where over 85% of electricity comes from coal, natural gas, and uranium. The inescapable fact is that hydrocarbons utterly dominate the information-communications-technology (ICT) energy supply chain where coal is, on average, the biggest player supplying 40% of domestic electricity. Just ask Greenpeace, which released its new analysis of Internet energy use earlier this year.
Greenpeace goes on to observe that the Internet's "electricity demand is expected to increase by 60% or more by 2020." No country, not even China, has that kind of growth. What worries Greenpeace is that "the internet's growing energy footprint has thus far been mostly concentrated in places where energy is the dirtiest." The word "dirty" in the Greenpeace lexicon means using coal as well as natural gas. Datacenter companies are, Greenpeace states, "choosing how to power their infrastructure based solely on lowest electricity prices."
Greenpeace is right about the primacy of price. Consider: Operating a single enterprise-class datacenter in a region with a low-cost coal-centric grid saves $350 million over the facility's lifespan, compared to operating in a high-cost region like California or New York. Global surveys of datacenter operators consistently reveal that cost and availability of electricity are the primary concerns.
Thus, for example, Google and Microsoft have built massive data centers in Iowa where coal supplies 70% of the electricity. As it happens, Iowa is also second only to Texas in wind generation. In Iowa, datacenter operators enjoy the benefit of saving a ton of cash, and getting green credit for investing a share of those savings in financing wind farms.
What about those wind turbines, you may ask. Well, wind can't supply datacenters directly because the Internet runs 24x7 and wind is sporadic. Big hydro dams are an alternative, of course, but there are precious few of those and they frequently generate environmental opposition too. Natural gas is an emerging alternative (in the U.S.), but its price is more volatile and presents a challenge for locking-in guaranteed savings for big electric users, and for many environmentalists shale gas is no more favored than coal.
The pressure on datacenter operators to find cheap reliable power is going to escalate. Chasing expensive power won't help. As the cost of data-centric technologies continues to decline, while electricity rates continue to climb, it will soon cost datacenter owners more to power computers than to buy them. The two costs are already roughly equal.
Meanwhile, demand for bits is escalating not just with the emergence of new big data services in developed nations, but also with torrid rise in data traffic in emerging economies. It is easy to see the physical linkage, not just the logical linkage, between the growth in electric-fueled bits and the source of electricity. According to the International Energy Agency (IEA), during the last 10 years of rapid tech growth, coal supplied 68 percent of the world's additional electricity supply, and will account for at least half going forward even if a forecast $4 trillion in global alternative energy subsidies are actually spent.
The global perspective matters, both for Stanford's principles and economic reality. The well-known tech giants own or use datacenters all over the world. And while they are the digital economy's icons, there are thousands of other lesser-known companies and tens of thousands of datacenters around the world, from Manhattan to Beijing, chasing the same cost metrics.
Speaking of China, for the sake of consistency, Stanford should embargo its portfolio managers from buying into the much anticipated and hot IPO of Alibaba which has an 80% market share of all on-line retail in China - a kind of eBay and Amazon combined, only bigger. Alibaba's massive datacenters operate, at low cost, on China's 80% coal-fired grid.
Stanford's experts doubtless know that their divestment, even if every other university endowment joined them (very few have), will have near zero impact on global coal consumption: mines won't close, power plants won't shut down, and datacenters won't go dark. In fact, data traffic will soar and datacenters will get built. But they may be increasingly built in countries where grids are cheap and reliable.
China is, for example, building out Cloud infrastructure at four times the rate of growth in its own domestic Internet services. In fact, China's Cloud is being built out faster than data demand growth in the entire Asia-Pacific region. Chinese planners have been candid about targeting global markets for their datacenters; their chief competitive advantage is not cheap labor, software, or better security, but low cost.
Big data is an ascendant trend in the world. Big data means big infrastructure. Big datacenters and big wireless networks consume big power. And in time-frames that matter, a lot of that power will come from coal because, in the end and always, price matters. The world's citizens will benefit from anything and everything that can facilitate data growth.
If Stanford's endowment wishes that someday coal will play a much smaller role, their focus would be more productively served in funding radical research hoping that one day some future Stanford scientist will invent an entirely new form of energy. But meanwhile, in today's world, growth will come, electric demand will rise, and economic principles inevitably dictate a major role for coal.