Is Chia really as green as itd marketing itself?

Hi,

I am currently doing my bachelors thesis about the environmental impact of different consensus mechanisms and blockchains.

When studying Chia I noticed that, although the current yearly energy consumption (while the market cap is still quite small compared to bigger ones) is around 0.3TWh (Eth 90TWh, Btc 200TWh) one issue with Chia and proof-of-space that I found when studying it is that it is driving up the sales of HDDs (Seagate reported +240% sales growth in April 2021). A study made by European Commission states that a HDD in server use (which is kinda similar to Chia) has an average lifespan of 4 years and only 6,3% of it can be recycled, mainly aluminium but not the actually critical materials. Also manufacturing a single HDD consumes 2KWh of energy.

Considering that currently Chia would require around 120k 10TB hard drives and if it gains more popularity these numbers would only go up, I have had a hard time to believe it would be an actually green blockchain like its trying to market itself. Its an improvement to proof-of-work for sure, but has a lot more environmental issues when compared to proof-of-stake.

Any thoughts?

one issue with Chia and proof-of-space that I found when studying it is that it is driving up the sales of HDDs

The HDDs retail price is transitory, just like GPU prices. GPU prices are over double what they were a couple years ago. HDDs are higher partially because of the initial gold rush for Chia but have started come back down. The same can’t really be said for GPUs. Nvidia & AMD recorded increase of total number units sold in 2021 vs 2020 but inflated prices have remained unchanged for over a year. HDDs on the other hand have already started to come back down to pre-rush prices after about 8 months.

A study made by European Commission states that a HDD in server use (which is kinda similar to Chia) has an average lifespan of 4 years and only 6,3% of it can be recycled, mainly aluminium but not the actually critical materials.

Chia farming is not anywhere near real enterprise server level use on a drive. I work in IT and have worked around data centers for years. The work load on a HDD for farming Chia is maybe 10% of what you would see in an active enterprise system. I would review the report you are referencing again to determine if the life span they are estimating is based on actual failure rate or frequency of replacement. Backblaze puts out a quarterly report of drives on their servers, many have a failure rate of less than 1% after 5 years.

Businesses operating enterprise systems will regularly replace drives when they reach the end of their warranty period, whether they are failing or not, because the cost of replacement is less than the business cost for data loss or unplanned system down time. So a drive that has a 3yr manufacturer warranty even without having any problems will be replaced in 3 yrs but possibly could still be operational well beyond that. The waste from business come from the outdated practice of destroying used drives due to security concerns. The Chia team has already started partnering with manufacturers to change this practice and create a secondary market for used drives that will reduce that waste.

As I said Chia farming does not have anywhere the same work load as an enterprise system, it is more like running a large local NAS. In the use case for Chia even consumer drives can last well beyond 4 years as long as they are properly managed. And Chia users would not have any reason to replace a drive unless there is an actual failure. Just like with PoW mining, users will run hardware until it breaks or simply isn’t capable of the work required, in Chia’s case as long as a drive is operational and is larger than a single plot size it is possible to farm from it. Drives used to store drives do not have the obsolescence of GPUs that are limited by the onboard memory to function correctly for mining/farming.

Also manufacturing a single HDD consumes 2KWh of energy.

And how much energy is consumed producing a single GPU? I dont personally know but considering they higher number of individual components and materials i would guess a GPU, especially a high end GPU, will at least be the same if not more to manufacture.

I have had a hard time to believe it would be an actually green blockchain like its trying to market itself. Its an improvement to proof-of-work for sure, but has a lot more environmental issues when compared to

In terms of “green” blockchain, consider the largest region that is ming or farming cryptocurrencies, which is China. About 62% of their energy production come from fossil fuels, mainly coal. Chia now consumes about 0.15% of what BTC mining consumes. Even if Chia grows to consume 10% of current BTC consumption, or 66x current netspace which would be possible but right now seems to be highly unlikely to occur, that is still a 90% reduction in energy consumption for a blockchain economy in a country that currently produces over 1/4 of global green house gas emissions.

Combine the much lower energy demand and options for reducing e-waste, it is a start to Chia living up to its “green” blockchain marketing. But add to that, Chia as a company has focused on real world uses for the blockchain to manage international carbon credits to create a public ledger for global carbon output that would make countries accountable for their envonmental impact.

All of that is my way of saying, Yes, I do think Chia so far is living up to its “green” ambition.

PS: I should comment on my used of quotes on “green”. No blockchain is really truly green, there is always some amount of energy and resources consumed in the process but the effort to reduce that consumption is what is important and the Chia team is trying to provide the best best balance of reduced energy and material cost while maintaining a highly decentralized blockchain.

Thank you for your very detailed post! I hope you don’t mind if I use some of these points you have brought up in my bachelors thesis? I need to find a reliable source that states that Chia works more like NAS than an enterprise server. The EU study if you want to check it out (its a long one) is here:

https://publications.jrc.ec.europa.eu/repository/handle/JRC96944

I should state again that after studying Chia I think it is a great improvement over PoW. HDD:s even according to that study there are much less harmful to environment than GPU’s or CPU’s. There’s also almost no possibility of Chia ever consuming as much of energy as PoW blockchains. However, in my opinion, Chia pales in comparison to PoS.

1. TPS. With the current block size and block time Chia is estimated to achieve 20 TPS. This can be scaled by implementing layer-2, but the throughput is still quite low compared to what has been achieved in PoS.

2. Energy consumption. PoS energy consumption basically depends on the number of nodes. Also, some PoS blockchain nodes can be run with very low energy consumption nodes (like RaspB Pi). Of course this can result in a problem where to run it as efficiently as possible the decentralization of the blockchain will suffer.

3. No requirement for specialized hardware. PoS, if nodes can be effficiently ran on any hardware, doesn’t require any specialized hardware.

On my thesis it is estimated that Cardano right now consumes 6 GWh per year and Algorand 4,9 GWh per year. Cardano TPS: 257, Algorand ~1300. There are of course weaknesses in PoS when it comes to decentralization and they way the consensus mechanism kinda stacks more currency to people with a lot of currency already. But from a purely environmental standpoint it seems to be the current leader.

Source: https://www.reddit.com/r/chia/comments/rcysq1/is_chia_really_as_green_as_itd_marketing_itself/