Since the cost and the environmental impact of a construction project don't matter anymore as long as Barack Obama is for it, here's a green infrastructure plan expensive enough even for Paul Krugman's taste: The North American Water and Power Alliance (NAWAPA):

Let's generate huge amounts of carbon-free hydroelectric power and turn the western half of the USA green (in the literal sense) by diverting a small fraction of the water from Canada's vast but almost-useless rivers that flow north into the Arctic Ocean. While we're at it, we can turn northern Mexico green, too. (Here's a promotional video from the 1960s.) Thayer Watkins writes:

The North American Water and Power Alliance (NAWAPA) is a project for diverting to the western U.S. and northwestern Mexico water from rivers in Alaska and Canada which now flow into the Arctic Ocean. In addition to providing irrigation water to arid parts of North America NAWAPA would also generate considerable amounts of power and provide some subsidiary benefits such as stabilizing the level of the Great Lakes. The project was formulated by the Los Angeles engineering firm of Ralph M. Parsons Company and got some attention in Congress, particularly from Senator Frank Moss of Utah, but is not politically feasible.

Canadians were outraged that Americans were planning to steal their precious bodily fluids, and were angry when George W. Bush mused about buying water from Canada in 2001. But, that's POT (Pre-Obama Thinking). We don't need that kind of negativism anymore, now that we have Obamagic.

In terms of engineering the project is feasible. A series of dams on the headwaters of the Yukon, Copper, Kootenay, Fraser, Peace, and Columbia Rivers can divert their flows into reservoirs. Included among these is the 500 mile long Rocky Mountain Trench, a natural formation which has 16 times the capacity of Lake Mead on the Colorado River. From the Rocky Mountain Trench the water would flow into Montana and central Idaho. The dams would generate electrical power but not all of it would be marketable. Some of the power would be required to pump the water over some mountains in Idaho to a canal where it would flow south along the border area of Utah and Nevada. Here the water flow would be divided into two branches. One would go southwest to Nevada, California, and northwestern Mexico. The other would go east to Arizona, New Mexico, and Colorado. This is the main element of the project. A subsidiary part would take water from the Peace River by canal to the Great Lakes and thereby linkthe praire provinces of Canada with the St. Lawrence Seaway. Other subsidiary elements could link the system to the Pacific Ocean at Vancouver, British Columbia and link Lake Manitoba to the Hudson Bay.

As envisioned by the R. M. Parsons Co. the system would deliver 120 million acre-feet of water annually; 78 million to the U.S., 22 million to Canada, and 20 million to Mexico. According to Parsons this would enable Mexico to triple her irrigated acreage, irrigate an additional 40 million acres in the U.S. and 7 million in Canada. NAWAPA would generate 70 million kilowatts of power; 38 million for the U.S., 30 million for Canada and 2 million for Mexico. Parsons estimates that all this would cost $100 billion in 1964 dollars. In 1989 dollars that would be about $339 billion.

Heck, Paulson and Bernanke can burn through $339 billion in just a few of their wild weekends.

Stanford computer scientist John McCarthy wrote about a decade ago:

We won't need any such grand projects for the forseeable future, but when and if our descendants need enormous increases in water supply, they can get them, perhaps at expense comparable in relation to per capita GDP to the expense our immediate ancestors spent on water projects. Probably the expense in proportion to the GDP of the region benefitted will not be as great as the 1904 Owens Valley aqueduct was in proportion to the GDP of Los Angeles at the time.

At that time, the population of Los Angeles was 200,000 and the per capita income for the U.S. was $1100. The cost of the project was $23 million. Therefore, it corresponded to 1/10 th of a year's income for the inhabitants of the area. 1/10 th of a year's GDP for the U.S. would come to $800 billion. It doesn't look like we will have to spend that much for increased water supply in the near future, but we'll do it if we have to. ...

Around 1900 people thought in large terms. Recently, it has become fashionable to think small.