Ask HN: Why is cloud storage so cheep?

Today Dropbox dropped prices. They now offer 1 TB for $10/month. Other providers like Google Drive have similar offers for quite some time.

I wonder how they do this. Dropbox uses Amazon S3 to save files. The lowest possible price on S3 is $0.0275/GB/month. If a Dropbox user utilizes all available space it would cost Dropbox $27.50/month. In addition most users are not paying for Dropbox but use there free offering adding files (=costs) to Dropbox. Running the company, development of the application, application server, download traffic etc. will come on top.

I see the following reasons why Dropbox might be able to make profit in the end:

- Most people are using much less space than the maximum of 1 TB. - Dropbox is doing deduplication. If two users have the same file Dropbox only needs to save it once. - Dropbox might have a better deal with Amazon and is not paying $0.0275/GB/month.

Do I overlook something? What is most important reason why Dropbox and others are able to make this prices?

I'm sure owning a large market share of a service storing consumer data is worth something, as unethical as that sounds... Abuse is fairly common as we've seen in recent years. When google scans a file into it's own format, or when OCR's run on your images to index information, who else can access these indexes?

A huge part of it is the practice of overselling. The basic idea is that you have n customers who all want k gigabytes of space, but on average they each use only j gigabytes of the space. If that holds, then rather than having nk gigabytes available, you just need n(k/j) gigabytes to satsify their needs. After amortizing usage, the people who overbuy and barely use their space more than makes up for the users that max out their space. I don't know for sure if Dropbox does this, but most hosting companies do, and its a sensible practice (especially when your backbone is scalable).

Pairing with that, nobody actually uses all of their space. If you see yourself getting within some percentage of your usage, one generally will buy more to keep from hitting that upper bound. Nobody likes being in the dangerous red zone.

I believe dropbox also compresses your data serverside, so you really use up n*p gigabytes (where n is your local usage, and p is the compression ratio), but don't quote me on that.

I think you have it there.

A terabyte is a nice round number, but I think getting a TB into cloud storage would be challenging, since most people have lower upload than download allowances.

At 1 Mbit/sec upload, for instance, it would take 8000 sec to upload a GB or 8 million sec to upload a TB, which amounts to a few months of continuous uploading.

That's a serious project for most people.

As for the other factors you'd better believe that they use compression, deduplication, special deals and every other trick in the book to bring down costs.

On top of all that, Dropbox has received over $1B in funding so it is very possible that they are losing money to build market share.

So far as cost analysis, I'd also look at the networking costs with AWS, since those are going to be significant. AWS doesn't charge for ingress, but it does charge about 5 cents to move a GB out at high volumes, and that can add up if a lot of stuff is being moved out.

So there's a couple things going on that would make this feasible.

1) People are unlikely to use that much actual storage in the first place. The tipping point, assuming their on S3 and haven't moved to a cheaper option like some other comments have suggested, for the $10/mon is ~363 GB. That's a lot of data for a single person to upload especially considering the poor state of US internet speeds.

2) There are technologies to push that break even point even higher from the customers perspective. Deduplication and compression will shrink many of a users files while still increasing their used count. It's really hard to estimate the effects of dedup and compression though because of the number of unknowns.