[BinaryIdiot]

Do we know how much emissions can be attributed to Amazon / AWS? I'm curious if this $10 billion fund could offset that or if it's too small to cover their own use.

Though, to be fair to Amazon, this is at least a good goal:

> Amazon expects 80% of its energy use to come from renewable energy sources by 2024, up from a current rate of 40%, before transition to zero emissions by 2030.

[Symmetry]

It all depends on whether you're comparing to a counterfactual or not. If you compare the emissions caused by AWS to those emissions simply not happening then that's a pretty severe source of carbon. If you compare it to companies hosting all that computing themselves then I think it's a pretty clear economic win. If you think that a lot of that would be self-hosted but many of the uses simply wouldn't exist then it's hard to say.

For Amazon I really feel like I have no idea about it. With warehouses instead of big box stores you've got a lot less volume being lighted and heated plus a lot of individual cars driving to a central place are replaced by a few large delivery fans then that's the same sort of win you get from people switching to driving the bus. But two-day shipping for rare items that involves flying them across the country is terrible for the environment. So I don't think that I have any clue what the net environmental impact of Amazon has been.

[wUabkSG6L5Bfa5]

SO MUCH THIS. People are always so quick to climb up Jeff's colon without bothering to compare to the status quo.

[jordigh]

> Do we know how much emissions can be attributed to Amazon / AWS?

No, the CDP routinely gives Amazon an F grade because they don't disclose anything:

https://www.cdp.net/en/responses/658?queries%255Bname%255D%3...

The purpose of the CDP is precisely for this kind of public shaming or praising, for keeping track of which companies are reporting what. Amazon is really bad for not disclosing anything.

Compare with Google:

https://www.cdp.net/en/responses?utf8=%E2%9C%93&queries%5Bna...

You can even read Google's reports here (I ignore most of the fluff and go straight to the appendix that has the actual numbers):

https://services.google.com/fh/files/misc/google_2019-enviro...

[natalyarostova]

Attributing emissions to Amazon/AWS is not a trivial problem. What you'd really want to estimate is AmazonEmissions-CounterfactualEmissions, which is the emissions caused by companies that would jointly provide the services demanded by consumers from Amazon, in the counterfactual world where Amazon didn't exist.

[jordigh]

It's not a trivial problem, but the GHG reporting protocol is widely understood and used:

https://en.wikipedia.org/wiki/Carbon_emissions_reporting#Gre...

If Amazon reported their emissions, we can corroborate their Scope 1 emissions with other companies' Scope 3 emissions and get an indication of just how much Amazon is directly responsible for.

Amazon, however, discloses nothing. This is the kind of thing that regulations are for. In other countries, disclosure is mandatory, and I would say for our survival as a species, absolutely essential. We need data before we can know what direction to take.

When we do our GHG assessment at the company where I work, we estimate our Scope 3 emissions based on what has been leaked about the location of Amazon's data centres and what we know about the electricity grid in those locations. We look at our CPU usage on the AWS console and estimate the energy required with some assumptions of the hardware.

It introduces a lot of uncertainty and we could do better if Amazon disclosed their emissions.

[standardUser]

It's kind of irrelevant how much harm a single company does, since that company exists to serve a demand that would simply be filled by other companies if it didn't exist. Maybe those theoretical other companies would have marginally better business practices vis-a-vis the environment, but it's not like they would operate in a wholly alternative economic system.

[maximente]

simplifying down complex issues to fit "the curves" narrative often misses a lot of really important nuance.

to wit, let's imagine there's an incumbent company that serves out the demand in an incredibly efficient but environmentally harmful manner. further suppose this is a large national player that's willing to eat some losses in order to crowd out the whole "efficient allocation of resources" type new businesses that may spring up and serve that demand in an environmentally friendly manner.

according to "the curves", there's no room to displace that incumbent because they're stuck in a local maximum when it comes exclusively to satisfying demand. but it's pretty clear that there are externalities that, when considered, make it an attractive target for change by different players.

[jordigh]

It's not irrelevant: a single company can undertake different economic activities that produce the same economic outputs with different ecological results. I've seen this first-hand in my line of work. When you make companies publish their greenhouse gas KPIs, they start to reduce them.

[throwawayhhakdl]

Project Vesta claims it would need $250B / year to sequester all the carbon we emit... https://projectvesta.org/