Market Research on market size and growth

Existing Markets

This post has some stats and ideas about the size of various similar markets that might use SAFE at various stages of adoption.

It’d be good to aggregate a bunch of stats into this topic if we can.

What’s the point? I guess it gives some indication of what success looks like, and promotes the potential scope of the network if it works as intended (both in size and purpose). It’s easy to see how revolutionary SAFE could be, or on the other hand how much work is still to come if the network becomes popular. These stats hopefully add something tangible to the overall concept of growth.

Early adopters

Debian torrent seeders

Popular linux distribution which can be downloaded from torrents shared by the debian community.

Total 18,431 seeders (sort of similar to vault operators). (source)

Most seeders for a single file is 849 seeders (seeding debian-9.5.0-amd64-DVD-1.iso).


2M connected users (source)

6K relays, which is like a routing-only vault (source)

275 Gbps available of which 125 Gbps is consumed (source)

100K onion addresses, which is like a SAFE site (source)


Distributed social network

623K users and 500 instances (which are like vaults) (source)


Proof of storage coin

Currently 288,000 TB being used for mining. (source)

Approx 10K installations of the android app gives some idea of non-mining adoption. (source)

Midterm adopters


20M users (source)


1M android app installs (source)


10M peers (source)

Dream Market

Dark net marketplace

26K registered forum users (from .onion site)

145K listings (from .onion site)


64M visitors per day (source)

Hacker News

Approx 500K registered users (from own data)


27M users (source)

Late adopters


130M subscribers reaching 300M people (source)


Approx 500M installs of the android app (source)


250M daily average pageviews (source)

140M daily unique devices (source)


330M users (source)


175M buyers and 25M sellers (source)

Windows Updates

200M updates to Windows 10 (source) at approx 3 GB per update (source)

To put that into context, 14M devices were running Windows 10 within 24h of release, which at 3 GB per update is over 4000 Gbps average bandwidth! Pretty serious infrastructure just for software updates. Think about android, ios, linux servers … software updates could be a big use case which is essentially invisible to the end users so lots of incentives for software providers to be switching to SAFE in that case.

Got some stats? Post them here!


And as soon as SAFE gets an accepted version of social media then what?

Great collection and analysis there @mav



Over 1 billion daily users.

Every 60 seconds:

  • 317,000 status updates
  • 400 new users
  • 147,000 photos uploaded
  • 54,000 shared links

Everyone on Facebook is no more than 3.5 degrees separated.

Source. It has a lot of other potentially interesting data too.


Not a stat, but I would put the pirate bay and torrent sites like it in the early adopter section. Especially private trackers. I think it will be very quick to pick up on the benefits.

Current procedure for any heavy torrenter:

  • Download torrent file
  • Upload torrent file to seedbox
  • Download torrent to seedbox
  • Download torrent files to local computer/nas
  • (When you’ve met ratio) delete files from seedbox

Procedure if files are are safe:

  • Download data map
  • (If you so desire a local copy) download to local computer/nas
  • Enjoy

I can imagine this extra step in the procedure for users who download movie torrent files: look for subtitles in own language (.srt file).

1 Like

The torrent sites is why I see a Patreon-like app in the SAFE Network as being a huge early app. It would allow torrent seeders to crowdsource funding for uploads (and fund their activities).


Another list of groups that may end up using SAFE.

These lists are useful for anyone thinking of writing an export/import tool, or for creating targeted advertising, or looking to build an app for a specific target audience.

Early adopters


Distributed marketplace

12K total users with 3K sellers and 35K items (source)


Self hosted cloud services

25M users (source)

That’s a huge number of users who are probably open to the idea of SAFE. Time to build a plugin?


P2P exchange

2.5K community members (source)

1K trades per month (source)

Midterm adopters


Privacy-focused email service

1M+ android installs (source)

Google Earth Engine

Data analysis platform

90K stackexchange users (source)

1M twitter followers (source)


Scientific papers

200K requests per day (source)

80K twitter followers (source)

Has significant problems with retaining domain names.

Late adopters


Video sharing

80M creators (source)


Photo sharing

90M monthly users (source)


IoT-style thermostat

11M sold (source)


Looking a bit deeper into The Pirate Bay…

The current Pirate Bay top 100 torrents accounts for about 176 GB of data and 450 TB of total data transfer[1] (this seems roughly consistent with the historical stats).

Note this is extremely simplified since it only counts currently active seeders and leechers. Many people don’t seed after leeching so are not represented in these statistics. I’m not sure how much that would increase the total data transfer, but presumably quite a lot, maybe up to 100x more?

The most popular torrent has more than 25,000 seeders so that’s perhaps a ballpark milestone to aim for in network size.

Anyway, this is gives some rough idea of what the usage of a popular file sharing service looks like and maybe allows some metrics for SAFE network to measure against for progress.

Overall this is a fair bit smaller than I’d have expected, but I guess it’s only top 100 so that’s not really much data in the scheme of things.

[1] SUM(Size * (Active Seeders + Leechers))
Ideally it should be SUM(Size * (Active + Inactive Seeders + Leechers))


porn is so efficient. We just reuse the same 176 GB over and over. I bet that small amount of data has resulted in Olympic swimming pools of pleasure.

In this case you only download it once then watch it off your hard drive. If it was safe network porn I would watch it on any device fearlessly because I would just access it again right off the safe network “drive” each time. This would result in alot more data transfer I wager. Same amount of initial porn though :stuck_out_tongue:


For big farmer would be worth to maintain own pornsite for free with most wanted content to rise overall download and get more GETs :slight_smile:


Burstcoin recently doubled mining capacity up from 300 TB to 600 PB


Can we use burst based network ?

1 Like

Filecoin went live in the last day or two, some crazy stats already!!

Network Storage: 581 PB

Top miner power:
40 PB
36 PB
24 PB
22 PB
20 PB

Miners on chain: 548 says slightly different stats

638 active miners

Mining geographical distribution
50% Asia
38% Europe
10% USA


10.911 FIL reward per block ($58 per FIL and 30s blocks is about $1.8M mined per day)

Block height is 151121 just for historical reference

Lots of links to charts and blockexplorers here:

And the mining requirements are pretty steep:

  • A miner will need an 8+ core CPU .
  • 128 GiB of RAM are needed at the very least. This should be complemented with 256 GiB of swap on a very fast NVMe SSD storage medium.
  • A powerful GPU is recommended as it can significantly speed up SNARK computations. See below for operations which can take advantange of GPU presence
  • a minimal amount of 1TiB NVMe-based disk space for cache storage is recommended. says costs to store is currently around $65/TB/year with no redundancy.

Some cost stats from

Cost per GB per year: $0.177 USD
35.88% Cheaper Filecoin vs Amazon S3 Standard
205.47 Days avg deal period (last 2.5K deals)
3,619.402 GB stored across the last 2.5K deals
8.339 FIL paid to store 3,619.402 GB (last 2.5K deals)


Interesting ycombinator thread. There is a lot of talk of censorship resistance, but given the lack of privacy, surely people will self censor?


FIL launch is not going so well apparently


Oh that sounds horrific, there are such big numbers involved in the filecoin project. I hope they can do something to mitigate that as we need competition in this space. I am interested in proof of space as it’s very complex and we have gone for a much simpler system of proof of resource. In our POS nodes are randomly asked for data or to check data (cryptographically and simply). So we went for don’t care how, but you better have the chunks and be able to deliver them. This bypasses the complexity of rsa vector proofs and the gammit of math.

Not saying we are better, we don’t know yet, but it would be neat to contrast and compare a math proven space claim to a “don’t care, you better have the data” claim.

In any case I hope they get around this and it does not become a huge corporate only network.



I can’t get my head round why they would be pitched at that level of hardware… do we know what a typical bitcoin node demands now?.. Safe’s intent for hardware seems a lot more realistic. That kind of RAM strikes like almost brute forcing the problem.


I think it’s for the data ‘sealing’ (proving you have a unique copy).

2GB RAM, 350GB disk space


Woah, post 13/n says: Total network capacity is currently around 600 PiB, for 4200PiB installed, or $240M worth of hardware.

That’s a lot of hardware!

Part of it is to do with the way proof of storage works. Burstcoin, chia,net, filecoin, all have this problem where space can be replaced by time (eg 1 TB of space can look like 10 TB of space if you have a powerful enough computer). So they all do proof of space-time rather than just proof of space. This is why verifiable delay functions (VDF) are such a hot topic right now, they feed into the time aspect of proof of space-time. It’s usually called a grinding attack (more info at the chia vdf competition).

I’m not personally so interested in VDF itself, usually it’s based on a repeated squarings algorithm, but the reason for VDF existing in the first place is quite fascinating and has a lot of implications for how energy efficient networks will function.