Here are some handy stats on power consumption in data centers Data Center Energy Summit taking place at Sun Microsystems today. Casually drop these factoids at your next cocktail party. Less than nine months: that the time it takes to recover the cost (in terms of lower electrical bills) for putting variable speed fans in a server rather than a fan that runs at a constant, high rate, according to Mukesh Kattar, energy director at Oracle. He tested it and thought payback would take 16 months, The fans, though, only had to operate at 55 percent of their regular speed on average, he said. 1 to 1.8. That's the ratio of power dedicated to support (air conditioning and cooling) compared to the power dedicated to actually running the server when you install these kinds of fans, Kattar added. In a regular situation, the ratio is 8 to 10. He's aiming for a 1 to 3 ratio. 80 degrees Fahrenheit. That's the temperature of the intake air servers can tolerate, according to Dean Nelson, who runs global lab and data center design services for Sun. Sun conducted a chill-off with several different cooling systems. Most servers are sold to work in environments with 65 degree intake air. (Intake air is the air that gets sucked in by fans to cool down the servers.). "So why are we cooling at 65," he asked. $19 billion. That's the total value of datacenter construction projects being undertaken by 21 large customers in the U.S. at the moment, according to Nelson. Data center construction is probably the only healthy segment of the building industry. A million bucks. That's what a raised floor costs in a good-sized data center, according to Nelson. Historically, raised floors were used to store the cables that connect computers as well as the cold air blowers. With new cooling techniques, you don't need these anymore. 50,000 square feet and 5 megawatts. That's the size and power consumption of a large data center today, according to Subodh Bapat, who runs Sun's energy efforts 500,000 square feet and 50 megawatts. That's the average large datacenter in 2020, he said. 8,500 hours. That's the number of hours in a year that you could cool your data center in San Francisco with ambient air. There are only 8,760 hours you have to worry about. $18.5 billion. The amount of money spent on data center power in 2005, according to Bapat. $250 billion. The amount of power spent on data center power if nothing is changed in 2012, according to numbers touted by Bapat from IDC. Gar.

Santa Clara, Calif.--Here's a novel idea. Some cities are examining the possibility of installing data centers, those energy-gobbling server rooms we all rely on, on mothballed ships. Although high-speed lines would have to be extended to the docks, the energy savings would be tremendous because these ship-bound data centers would need far less air conditioning than standard data centers. Air conditioning accounts for 33 to 50 percent of the power that gets pumped into data centers. The air is cool in many of the world's seaports. More important, ships sit in the water.  "It (a ship) has the biggest heat sink in the world beneath it," said Subodh Bapat, who oversees Sun's green and energy efficiency efforts. Sun is hosting a data center energy summit today. For the past two years, Sun has been touting the energy efficiency message hard.  In another place, a municipality is contemplating taking the hot water that gets produced in data centers and pumping it--after filtration--into its public pools. The water that comes from data centers never gets hot enough to crank turbines, but it's warm enough for applications such as this.  Ambient air cooling with air-side economizers is also gaining traction. In San Francisco, one study noted that ambient air is cool enough to cool server rooms in that city 8,500 hours a year. Cooling is only needed 8,760 hours a year, he said. Thus, anyone building data centers there can whack their electrical costs by designing the building with passive air cooling in mind. (Microsoft earlier this year told me that they were going to take advantage of ambient air cooling in a new data center going up in Ireland.) Ambient air can also be amplified with a technology called earth pipes. With earth pipes, air gets pumped into a miles-long network of pipes approximately 30 feet underground. The air gets chilled there and then comes up to cool off the server room. (Side note: the ancient Egyptians used a similar technology.) Even with the cost that comes with filtering the air of particles and moisture, ambient air is still cheaper than air conditioning, according to Bapat. Other companies are also implementing the ice cube concept. In this, cheap electricity at night is used to run ice makers. The ice then melts during the day: the cool air acts as air conditioning. Remember sticking your head in the freezer as a kid and breathing in the air? Same concept. And here's another idea: let computers tolerate warmer air, according to Dean Nelson, who runs the global lab and datacenter design services at the company. If they can live in warmer temperatures, less air conditioning is needed.       

Despite the near daily hum telling us that we're in big trouble, sometimes we read something that makes us think: wow, we're in big trouble. An Ernst & Young report released yesterday profiling 40 benchmark oil exploration and production companies in the U.S. found oil production has remained flat at 1.2 billion barrels per year since 2004, after declining from 1.3 billion barrels per year in 2003. The benchmark companies represent 74 percent of U.S. oil reserves, which have also remained flat since 2006 at 16.1 billion barrels. Actually, to break that down a little more, proven reserves from independent oil companies were actually up seven percent over the last year. Proven reserves from the integrated companies - Exxon, Chevron, etc. - were down two percent over the same period. But maybe things aren't as bad as they seem. Surely the oil companies must have cut production to reflect some other market variable, like bottlenecking through our nation's inadequate refining capacity. How else do you explain those margins? Actually, the oil companies are pretty much screwed. Exploration costs - the amount companies spend to find oil - increased 165 percent between 2003 and 2007 to $12.8 billion. Development costs - the amount companies spend to take petroleum out of the ground - grew by 180 percent to $18.4 billion over the same period. Growing demand pushed up revenue 12 percent to $141.5 billion on the year, despite the fact that the price per barrel actually declined slightly from January 1, 2006 to January 1, 2007. Significantly, however, net revenue increased only four percent "due to rising production costs and increases in depletion, depreciation, and amortization." The cost of finding and extracting oil is growing at an obscene rate, while revenues have been propped up by high demand. As prices rise to reflect market tightness, we can expect income to continue falling as exploration and production costs continue their meteoric rise. Falling demand, at least in the U.S., can't be far behind. Someplace, somewhere Denis Hayes is smiling. Actually, it's right over here. As Hayes takes the latest cap-and-trade bill to die a miserable death in the Senate to task, he also outlines an effective policy proposal that may find some support among U.S. oil exploration and production companies. Hayes argues "the backbone of any comprehensive policy to limit greenhouse gas emissions must cap carbon at the places - coal mines, oil fields, pipelines, ports - where it enters the economy." The proposals in Lieberman-Warner, as with those about to come into force as part of RGGI and currently in force in the European Union, place a cap on carbon at the points where it enters the atmosphere. Regulating the latter would proved to be nearly impossible. In 2006 there were 336,000 factories in the United States, more than 10,000 coal, natural gas, and petroleum generators, and roughly 150 oil refineries. By comparison, Hayes estimates there are 2,000 points in the U.S. where carbon dioxide enters the economy. Given that oil production has remained flat since 2004, and is likely declining based on historical trends, it may actually be to the benefit of oil companies to hop on board with Hayes's proposal. With the number of point sources for carbon dioxide entering the economy likely to fall, oil producers would feel the pinch less in the long-term than if the suddenly saw a policy-influenced demand drop as factories are forced to switch to cleaner power sources or it automakers are forced to internalize the carbon cost of the cars they produce. Furthermore, capping carbon dioxide at its emissions source would serve to regulate the amount of oil or natural gas or coal allowed to enter the economy. This would let these companies internally regulate their production to match their carbon allowances, extending out their shrinking supply while charging a much higher price. Not only would this save on further exploration costs - there's no need to develop heavy oil or bituminous sands if you know last year's exploration can stay in the ground for another year or two - but it may even help prop up that flagging net income rate.

Buying the most energy efficient server or desktop on the market costs more, but it's not outrageous, according to Albert Esser, Dell’s Vice President of Power and Infrastructure Solutions. "There is a premium but it is sensible. The ROI is about a year or less," he said. Esser was talking about a line of servers Dell is coming out with that sport a power supply that is 92 percent efficient. That means that 92 percent of the power injected into it gets used productively. Currently, Dell servers have power supplies that are around 80 percent efficient. Dell is also coming out with servers and desktops with power supplies that are, respectively, 89 and 88 percent efficient. Traditionally, server power supplies could be 60 to 80 percent efficient, he said. It doesn't sound like much, but if you add it up over millions of servers cranking out spam messages around the world, those electrons add up. Hats off again to the tech industry. Years ago, these companies didn't care about power bills. Intel and AMD began to research energy efficient chips at the start of the decade, but mostly to contain heat inside computers. Electrical costs weren't a big factor. But when power bills began to escalate in 2005, the obsessive, hypercompetitive men and women in the industry began to rapidly tackle energy consumption. There has probably been more tangible improvement in the IT industry as far as energy consumption is concerned than any other industry.

In the chip industry, the size of chips typically only goes one way: down. By exploiting the properties of Moore's Law, manufacturers can shrink the size of the transistors that go into their semiconductors every two years and get chips that take up less space. Although manufacturers might add features and transistors, the size of their devices almost invariably shrinks at a steady pace. (Shrinking also makes them cheaper.) Luminus Devices, the light emitting diode (LED) spin-out from MIT, however, has found that you can get interesting properties out of large chips. Rather than make small LEDs that might measure 1 millimeter a side and take up a little more than a square millimeter in area, the company makes devices like its PT120 that can sport 12 square millimeters of light emitting surface (that's 4.6 x 2.6 millimeters). A larger LED means that fewer LEDs are needed to produce a lamp, which in turns leads to higher efficiencies. "It ends up being more light per square millimeter," said founder and CTO Alexei Erchak. "You can't perfectly pack LEDs" into lamps. The light that comes out of these larger LEDs is then chanelled by photonic lattices. The lattices--think of them as nano-sized pieces of optical communications equipment--effectively take the light that would radiate in all directions and force it out through the surface of the chip. Other reports have talked about the lattice, but generally neglected the size of the chip. That large size sort of puts the pieces of the puzzle together. Laugh, but for some like me who's spent a good part of his adult life studying chip size, this is pretty cool. Several other researchers have tried to incorporate photonic lattices into products, but it hasn't panned out. Luminus, says Erchak, is the first company to bring out a product that commercially exploits them. "Photonic lattices were kind of overhyped in the late 90s," he said. "People thought they could be used in telecom." Like Luxim, Luminus started out making light sources for projectors and projection TVs. Now that projection TVs are going away, Luminus is moving toward making lights for LCD TV makers (it has a deal with contract manufacturer Jabil Circuit) and for commercial and architectural lighting. LEDs still cost more than conventional lights, but they require less maintenance and consume less power. Thus, retailers, municipalities and large commercial building owners are already scoping out high-powered LEDs for garages and public gathering places. (Both companies, by the way, landed millions in VC money in recent months. Lighting is hot.) The company's PhlatLight CBM-360, which will begin to come out in sample quantities in the third quarter, will mark its entry into the market for white commercial lights. The produce will sport a 36 square millimeter emitting and be capable of putting out 4,000 lumens. (It can also be tuned to emit 80 lumens per watt if energy efficiency is more paramount than output.). Either way, that's one honking bulb.

New Jersey's Public Service Electric & Gas Company, an investor owned utility, has filed a proposal with the state's Board of Public Utilities to launch a $45.9 million demand side management program. Demand side management (DSM) is a method utilities use to reduce consumer demand for electricity and gas through the introduction of efficiency measures and technologies. In regulatory environments where utilities are constrained by the amount of electricity they are allowed to produce, typically through mechanisms that determine ratemaking as a function of fixed asset investment alone and not through a combination of fixed asset investment and commodity sales, demand side management is used to curb consumer demand. Under the former regulatory mechanism, known commonly as decoupling, utilities are punished for exceeding their planned electricity production schedule. On the flip side, consumers - the ratepayers - are punished if utilities under produce and are unable to make a sufficient rate of return on their fixed asset investment. Such a regulatory mechanism requires a balance between consumer demand and utility supply, which can be accomplished through a combination of demand side management and integrated resource planning (IRP). DSM is the easier and less expensive of the two. While IRP may require the introduction of renewable generation capacity, DSM typically involves utilities helping consumers insulate home heating systems, passing out CFL bulbs, conducting home energy audits, or upgrading consumers to programmable thermostats. In other words, it targets the lowest of the low hanging fruit. And this is exactly what PSE&G is doing. What's interesting here is that New Jersey's utilities are regulated under the traditional regulatory design scheme, which was described in the early 1960s in a paper by Harvey Averch and Leland Johnson. The Averch-Johnson Hypothesis describes how utilities tend to overcapitalize due to incentives inherent in traditional rate of return regulation. Briefly, overcapitalization maximizes profits under rate of return regulation because the return on capital investment is linked to the amount of the commodity - electricity - that is produced and sold. More electricity requires more power plants. In unregulated environments, as any first year B-school could tell you, utilities would invest in and produce only what the market demanded. In other words, rate of return regulation distorts the supply market by encouraging over production. As a result, consumers tend to use more electricity than they would otherwise. Regulators encourage this activity by mandating rates remain low. So why would a utility in such a regulatory environment intentionally shoot its balance sheet in the foot? New Jersey is one of the signatories to the Regional Greenhouse Gas Initiative, a multi-state agreement that calls for a 10 percent reduction in greenhouse gas emissions below 1990 levels by 2018 for 10 New England and Mid Atlantic states. RGGI, at least initially, is targeted specifically at power utilities. To attain this goal, RGGI has introduced a cap-and-trade system that will have its first emissions permit auction in September 2008. While PSE&G has neglected to mention the motivation behind the DSM proposal, it is likely they are undertaking the program as a way of reducing its exposure to what may be a fairly pricey emissions trading market. By encouraging conservation and energy efficiency among its consumers, PSE&G will be able to produce less electricity, thus reducing their rate of carbon emissions and the number of permits they will need to buy. This effect is similar to what occurred in the early 1990s with the introduction of a cap-and-trade scheme aimed at mitigating acid rain, except those changes occurred on the supply side. Twenty years ago, utilities began using SOX and NOX scrubbing technology because it was cheaper than buying emissions permits. This is also the same, though again on the supply side, as the European utilities that have invested billions of euros in wind capacity. This leads to an interesting, though fairly obvious, conclusion. Utilities would rather make adjustments now that they know to be both cheaper and more controllable to their electricity production and greenhouse gas emissions rates than subject themselves later to a volatile and less certain market mechanism that they will be forced to compete in. Or at least introduce measures to hedge against potential future exposure. Though PSE&G's DSM proposal is fairly small scale it helps put in context actions undertaken by larger utilities - XCel's Smart Grid City in Colorado and Duke Energy's $100 million residential solar plan are both good examples, though these may also be viewed partly as measures aimed at complying with state-based renewable portfolio standards.

Presidential candidate John McCain unfurled part of a plan to cut down on oil consumption and imports. Some of it makes sense, but some of it is grounded in politics and, I think, a too-glossy view of how the transportation market works. Here are the highlights: 1. A $5,000 tax credit for people buying a zero emissions car. That works. It's a pretty hefty credit, but since there aren't many zero emissions electric cars out there, the Federal government won't find itself in a budget crunch for a few years. Still, it gives an incentive for companies to produce them and consumers to buy them. One of the reasons electric cars have failed to date to catch fire with consumers is the comparatively high price. 2. A $300 million prize for developing an electric car battery that leapfrog's current capabilities. A nice bonus for anyone who gets it, but probably a red herring. Contests are rapidly becoming what advertising was to late 90s Internet companies: an easy, if unlikely, answer to all of your problems. The DARPA Challenge helped jumpstart a push into robotic cars, but that's been the principle accomplishment. (Besides, the Department of Defense, which supports DARPA, is probably the primary consumer of robotic cars.) The Ansari X Prize allowed a man to get into space, but do you really think we will have space tourism any day soon? Venture money is already flooding into batteries. Researchers around the world are also bending their minds to the task. A123 Systems, Firefly and even Exxon Mobil are trying to improve batteries. In other words, large amounts of capital and know-how are already on the job. And to top it off, the Department of Energy doesn't have a great record for follow-through. Remember the hydrogen highway? The FutureGen clean coal project? The contest won't change it much, but it will look nice on paper. 3. Eliminate the effect of tariffs and subsidies on ethanol. Subsidies and tariffs distort market forces. True, but getting a new industry off the ground sometimes requires government support. Brazil subsidized ethanol for around 15 years before eliminating tariffs, and it eliminated tariffs because it could no longer afford them. Subsidies, in other words, may have to stick around for a bit to help get the U.S. industry off the ground. And even after Brazil got rid of the subsidies, the government continues to support the industry. Gas stations serve up fuel that consists of 20 to 24 percent ethanol. Oh, that darn market interference again. 4. Flex fuel cars will help solve the problem. General Motors produces lots of flex fuel vehicles now. Unfortunately, there are very few stations in the U.S. that pump ethanol. The total comes to around 1,400. We will need 15,000, according to GM. Unless more stations pop up, and the government mandates a larger percentage of ethanol in regular gas, flex fuel cars won't change the picture much 5. But, most important, his platform isn't entirely based around drilling for oil. Give him pointers for that.