commit 472821afd2da0104fb5c3ecd2e65f12909f27e85
parent ad5e0e80020882c546421478a39761c0e115a515
Author: Andrew Laack <andrew@laack.co>
Date: Sat, 27 Sep 2025 17:35:04 -0500
Continued work on yt post'
Diffstat:
2 files changed, 14 insertions(+), 10 deletions(-)
diff --git a/posts/sustainability-of-youtube.gmi b/posts/sustainability-of-youtube.gmi
@@ -10,21 +10,13 @@ Similarly, I am skeptical of 'free' services. It's incorrect to say "if somethin
When will YouTube's storage costs exceed their revenue if they don't start purging old content?
-## Hypothesis
-
-NOTE: My hypothesis is not well researched. This section exists because I think it is important to share intuitions and identify mistakes in reasoning.
-
-My inclination is the physical limits of the universe constraining information density are not the bottleneck for their data storage problems. I suspect the cost of maintaining (resilient) data centers for video storage will be the bottleneck. My unresearched guess is they have ~10PB of data uploaded and compressed onto their servers per day. I think I can buy 1TB of storage for ~$10 if I buy a larger HDD so it might cost me ~$100,000 to buy enough hardware the data uploaded to their servers in one day. I would also assume they get a substantial discount on their drives, but they also need redundancy and a location to keep them.
-
-This cost seems trivial compared to the compounding cost of maintaining large scale data storage so I don't think it should be taken into account for my hypothesis. If they need to replace drives on average every 6 years this amounts to ~$1,666,666/exabyte/year. I'd also think all supporting costs add up to approximately $2,000,000 per year. I suspect YouTube's max potential earnings are ~$300,000,000,000 per year, regardless of external factors. This leads me to believe YouTube may only be able to support ~81,818 exabytes of data. At my assumed current rate of 10PB saved to their servers per day we find they will be forced to start purging content from their service in 22,416 years. Man, I was not expecting such a high number... While this is what I conclude from my uninformed numerical guessing, if you were to ask me, I would guess their business would become unsustainable in ~50 years.
-
## Findings
YouTube states on their official blog there are over 20 million videos uploaded per day [4]. I can not find any believable metrics for how much data this is because so many of these SEO slop sites regurgitate the same numbers which don't correspond with YouTube's stated number of video uploads per day. While I don't trust YouTube very much, and they likely have incentives to inflate the numbers, they seem more trustworthy in this context as they are, in fact, the ones who are hosting the content. As such, I will accept this metric and try to work backwards to an approximate amount of data being saved to their servers.
### Scraping
-I wrote a simple python script [5] to scrape YouTube's 100 most recent videos based on the top 2,711 trending search terms on Google in the US on 9/13/2025 for the past 7 days [6].
+I wrote a simple python script [5] that used a curated list of popular Google Trends searches over the past few decades [6] that uses the YouTube search endpoint, sorting by most recently uploaded, to compile a list of ~7.65 million YouTube video URLs along with their duration.
While this seems to be a reasonable proxy for YouTube video lengths, there are some limitations of the approach such as:
@@ -45,4 +37,3 @@ These limitations, and more, are flaws in my methodology that make them not enti
=> https://en.wikipedia.org/wiki/Bekenstein_bound Theoretical limit for information storage in a finite region of space
=> https://web.archive.org/web/20250911091711/https://blog.youtube/press/ Videos uploaded per day
=> gemeini://blog.laack.co/python_script Python Script for YouTube
-=> https://trends.google.com/trending?geo=US&hl=en-US&hours=168 Trending Search Terms in the US
diff --git a/python/youtube/results/math.md b/python/youtube/results/math.md
@@ -0,0 +1,13 @@
+1.4447334047285894 GB/Video * 20,000,000 Videos/Day
+
+= ~28.895 PB Per Day (approx. upper bound)
+
+---
+
+396.17368260710157 MB / Video * 20,000,000 Videos / Day
+y
+= ~7.923 PB Per Day (approx. lower bound)
+
+---
+
+
