Retail giant Amazon has a rather mixed reputation among progressives, to say the least. There are ongoing debates about its labor practices and its ruthless competitive tactics, which have driven competitor after competitor out of business. Among environmentalists, opinion is also divided — some argue that online shopping (and shipping) has accelerated wasteful consumer culture, others that it has reduced the carbon footprint of shopping that was going to happen anyway.

Putting those broader disputes aside, there is one area where the company is making substantial and undeniable progress. After years of what could charitably be called benign neglect, Amazon has recently moved aggressively toward corporate sustainability.

It plans to have 15 rooftop solar systems, with a total capacity of around 41 MW, deployed atop fulfillment centers by the end of this year, with plans to have 50 such systems installed by 2020.

Amazon was the lead corporate purchaser of green energy in 2016. That year, it also announced its largest wind energy project to date, the 253 MW Amazon Wind Farm Texas. Overall, the company says, it has “announced or commenced construction on wind and solar projects that will generate a total of 3.6 million megawatt hours (MWh) of renewable energy annually.”

(Of particular interest to Vox nerds: The company also signed on to an amicus brief in support of Clean Power Plan, Obama’s signature climate policy, which the Trump administration wants to kill.)

More from Vox:
The world’s largest car market just announced an imminent end to gas and diesel car
How free porn enriched the tech industry — and ruined the lives of actors
Genetically engineered humans will arrive sooner than you think. And we’re not ready.

But what recently caught my eye is the clever way that Amazon’s giant new campus in downtown Seattle will keep its occupants warm in the winter — a pleasingly low-tech solution that sits squarely at the intersection of several of my enduring obsessions: the importance of smart design, the benefits of urban density, and the need for electrification of the power system.

Juicy! Let’s take a look.

Amazon is building a giant four-block campus in downtown Seattle’s Denny Triangle, which will eventually encompass five towers and five million square feet. This schematic is from Geekwire (which has its own excellent story on Amazon’s project).

That’s a lot of space to heat during the dreary Pacific Northwest winters.

Good fortune! Just across 6th Avenue from Doppler Tower is the 34-story Westin Building Exchange, a regional telecom “carrier hotel” that houses computer and server hardware for some 250 telecom and internet companies. About 70 percent of the building is occupied by data centers.

All that hardware creates an enormous amount of heat. Lacking any better options, Westin has been venting that heat into the Seattle air, through giant cooling towers, at considerable expense.

Waste heat, meet heat consumer.

Amazon and the co-owners of the Westin — Seattle’s Clise Properties and San Francisco’s Digital Realty Trust — struck a deal: Amazon would buy some of the Westin’s waste heat.

Tweet: As winter approaches, an innovative heat source will be keeping our Seattle headquarters warm. #greenenergyhttp://amzn.to/2j2t8pI

The technology is impressive at this scale, but conceptually, it’s pretty simple. Water circulates in PVC pipes throughout the Westin, gathering heat. It is then sent beneath the street (city waivers had to be granted to cross public rights of way, a huge problem for waste-heat projects historically) to Doppler tower.

Beneath Doppler, a series of large heat exchangers run the Westin’s warm water next to cool water from Amazon, heating it up (exchanging the heat). That warmed water is around 65 degrees, so it is run through a series of five heat-reclaiming chillers to raise the temperature to 130 degrees (also reducing the volume of water).

That hot water is then circulated beneath the floors of the campus buildings via radiant heating systems, providing more than three-quarters of the heating required by the campus, at about four times the energy efficiency of a comparable HVAC system.

The cooled water then returns to the basement to gather more heat, and lo, the cycle continues.

Westin’s building manager told the Seattle Times that the building generates the heat equivalent of about 11 megawatts. During winter, Amazon will siphon off about 5 megawatts of that.

In effect, Westin is replacing Amazon’s need for natural gas boilers to heat the water, which would have run close to a million dollars annually in fuel and maintenance costs. Another way of looking at it: the system will allow Amazon to avoid the purchase of some 4 million kilowatt-hours of electricity a year.

What’s more, the data center is a rock-solid, reliable source of heat; it is guaranteed to always be running (it has backup generators if the grid goes down). Amazon has a couple of efficient natural gas boilers for backup, but at least over the past two winters, it says they have rarely been run.

In exchange, during the winter, Amazon’s campus is serving as a giant cooling tower for the Westin, substantially reducing its cooling bill — and paying for the privilege.

It is, to risk cliché, a win-win.

To pull all this off, Amazon and Clise had to create an “ecodistrict” (navigating a thicket of city bureaucracy). The coolest part is that there’s lots more heat to use, so there’s no reason the district couldn’t be extended to more Amazon buildings, or to surrounding buildings.

There’s nothing new about “district heating” — i.e., heating multiple buildings using a single source and a network of pipes. The concept dates all the way back to the Roman Empire. In the Industrial Era, such systems were frequently used to exploit the heat from small in-city power plants. A number of modern cities, from Paris to Tokyo to Vancouver, BC, have district heating systems in use today, fueled by everything from biomass to sewer water to solid waste incineration.

What is somewhat novel about Amazon’s arrangement is the use of an electricity consumer — a data center — as the heat source. (Waste heat from data centers is used in other places, though Amazon says this is the first contract between a data center and a heat consumer with different owners.)

As I mentioned, this sits right at the intersection of three of my enduring obsessions.

First, it represents enormous energy savings with relatively old and reliable technology — it’s just water run through pipes. What produced the savings was not some whiz-bang tech innovation, but simple design. Cities produce enormous amounts of waste heat and then import enormous amounts of energy to warm themselves. It’s kinda crazy. Smarter design could help close that loop, reducing the need for imports, reducing emissions, and bringing more of a city’s energy management under its own control.

Second, along the same lines: cities! Cities are everything this century. State, provincial, and national governments are often captured by rural or corporate interests, creating substantial status quo bias. In cities, however, constant change is a fact of life and governments are at least somewhat more nimble — and that’s especially true when it comes to experimentation and innovation around the clean-energy transition. The more cities can generate and manage their own energy, the less hostage they are to state and national policy shifts.

With that in mind, why not bring more highly energy-intensive operations — not just data centers, but, oh, indoor farms, or marijuana grow operations — into dense urban areas? They have traditionally been built in the hinterlands, where land and power are cheap, but if they could make a substantial second income selling their waste heat, they might be able to justify urban locations.

That would reduce shipping supply lines. It would bring tax-paying, profit-making industrial facilities that don’t pollute into urban boundaries and budgets. It could potentially reduce urban HVAC bills, if the facilities are hooked up to district heating systems. And it would give cities more direct control over their energy fate.

Third: Amazon opted for a hydronic (fluid-circulating) system, drawing on local sources of heat — not direct electrical heating. Is this an exception to my “electrify everything” mantra?

Yes and no. Mostly no. The key difference here is that the local source of heat is itself electrical — a data center. So in that sense, Amazon is using electricity for heat. In fact, it is helping Seattle use its electricity more efficiently, getting more work out of the same kWh.

The expert community is divided on the possibility, or advisability, of moving all HVAC to electricity (like heat pumps). It would require an enormous amount of capital and produce an enormous new electricity load. Limitations on both probably mean that we’ll need zero-carbon HVAC alternatives that still draw on heated water or combusted fuel, which can substitute more easily in retrofits of today’s buildings.

Perhaps some of that fuel can itself come from electricity — it is increasingly possible to synthesize liquid fuels that way. But a great deal of the heated water can come from existing and planned sources of waste heat, preferably drawing on electrical sources.

If electricity is zero carbon, then all subsequent derivatives, from waste heat to synthetic liquid fuels, will be zero carbon as well. That is the magic of clean electricity.

There’s a telling line in Amazon’s blog post on this project:

Because of the up-front investment to build it, the ecodistrict isn’t currently delivering heat that’s any cheaper compared to a typical system, according to [Mike Moriarty, the Senior Engineering Manager who leads operations and maintenance for campus], but “the price of electricity is only going to go up.”

This is like so many clean-energy technologies: The upfront investment is high relative to the alternatives, but it pays back many times over throughout its life.

As Amazon adds buildings to the system — and maybe someday, as Seattle expands the system beyond Amazon’s campus — the savings will pile up ever higher, eventually dwarfing the upfront investment.

A robust district energy system could save Seattle money for a hundred years. Again, it’s just pipes with water, a technology that has stood the test of time. All that’s needed is policy coordination and good design.

Whatever its other sins, Amazon has shown itself willing to make patient investments; that’s why it keeps growing but never makes much profit. It invests for the long term.

Almost more than anything, it is that mentality — the willingness to sink upfront investment into long-term savings — that can move corporations (and society) toward sustainability. For that reason alone, I hope this project pays off for Amazon.

And I hope Amazon carries the lesson over to its next headquarters.