Why does Dropbox still use Amazon S3 to store data instead of building its own data center? Is it still cost effective for Dropbox?

As the CEO of an enterprise Cloud Storage/Cloud File Server service provider (DriveHQ), I’d like to offer my 2 cents on this.

When Dropbox was started in 2007, the cloud storage price was much higher. Amazon S3’s price was considerably lower. Also due to S3’s tiered pricing, and assuming Dropbox received a 40% discount, using S3 storage probably made sense for Dropbox.

However, S3 was never really cheap. When you factor in that Dropbox, DriveHQ and other cloud storage services usually offer storage space to free users, and that most of users are free, S3 pricing is (was) actually very expensive. So the main reason for Dropbox to choose S3 was not because of lower storage cost; it was because of time to market, lower initial capital investment and better scalability. (In the case of DriveHQ, we started in 2003 before S3 was launched. So we could not have used S3).

The Cloud Storage market has since changed dramatically. Most consumers are expecting free storage or unlimited storage, and most cloud storage providers had to lower their storage price again and again. However, S3’s price did not drop as much as other cloud storage services. Today, S3 pricing is often higher than other cloud storage services. Many people are misled to believe that S3 storage is cheap, which is not true. S3 has a complicated pricing model that charges much more than just storage space. Their bandwidth cost alone can be 4 times of storage price.

Clearly Dropbox cannot continue using S3, and indeed, they mainly stopped using S3 early last year.

Why did it take so long? I believe it was mainly caused by technical difficulties and the huge cost associated with this process. Dropbox had relied on Amazon S3 for many years, thus they lacked internal expertise/resource on managing such gigantic amount of data. Moreover, they had to make sure such migration would be fail-proof as they had hundreds of millions of users using their service! They did it, kudos to them!

Because that's best for the business and its shareholders due to simple math.

Gross profits at now + time t are current revenue * growth ^ t * gross margin where gross margin is ((revenue - cost of goods sold) / revenue)

Generally it's much better for startups to devote resources to increasing growth as the exponent base than reducing cost of goods sold which merely multiplies the effects of the exponentation with a limit somewhere below a 100% gross margin.

Forbes estimated Dropbox revenue on track to hit $240M in 2011 and nearly $500M in 2012 which is 108.3% year-over-year growth or 6.3% monthly.

At that growth rate they'll end 2015 with a $4.5B annual run rate.

At the same point in 2011 and1% lower growth rate or 5.3% monthly they'd have gotten to $446M in 2012 and end 2015 at $2.7B

At 1% more or 7.3% monthly their current revenues would have been $558M in 2012 and end 2015 at $7.1B.

It'd take a 37% cost reduction to increase profits the same amount by year-end 2015. Another year of growth makes it 74%. Three years out gross profit improvements from cost-of-goods-sold reduction are not mathematically possible.

Simple arithmetic aside, growth rate is also important as the mechanism which wins land grabs. DropBox is fighting Google Drive, Apple's iCloud, and potentially a post-Balmer Microsoft SkyDrive which plays nice outside their ecosystem.

Dropbox and other startups are usually better off dedicating engineering resources to features that increase growth instead of cutting costs of goods sold via things like a lower cost alternative to Amazon Web Services.

They're also usually better off spending money growing marketing to increase growth rate than engineering and/or operations to increase gross margin.

They're generally better off closing the feedback loop on riskier but potentially more significant improvements (as in bigger addressable markets and more organic growth from each customer) than the more bounded gains from cost of goods sold reduction. More years of higher growth will have a huge compounding effect and when they bet wrong they can try the next high-risk high-reward improvement.

At startups these things are even more important due to the time value of money. Early $10M can be half the company while later in life it should drop below the noise floor. Deferring things which don't need to be done for survival or to increase growth can allow less dilution with bigger wins for the founders, employees, and investors.

Once close enough to the natural limits of their markets without others to exploit and with the land grab won Dropbox will focus on reducing cost of goods sold for better gross margins and profits at which point data centers without Amazon's profits may win.

Update: that’s what they did, finishing in 2016 three years after I wrote my answer. The Epic Story of Dropbox's Exodus From the Amazon Cloud Empire

Even though this question dates back to 2013, I thought I'd write an update since it popped up as the 3rd result under the Amazon Web Services category.

Dropbox has switched away from Amazon S3 to store data, and this post does a great job of explaining it: The Epic Story of Dropbox’s Exodus From the Amazon Cloud Empire

TL;DR: Cost. Also, Dropbox is at a size where they believed they could pull this off without shooting themselves in the foot. I'm sure it took a lot of engineering power and skill. It is also a risky move, especially if their business doesn't continue to grow at the expected rate.

Thanks for the A2A.

As mentioned by Christophe Limpalair Dropbox no longer depends on AWS for its needs.

The advantage that cloud gives is that it helps corporations move fix capital cost to working capital. Imagine a corporation which is growing fast, Netflix is a great example, it will have to keep on investing in generic hardware according to revenue growth that it is expecting. By moving to cloud it is able to forego this capital investment.

The issue in this scenario is that AWS penalizes an entity if it breaches the data transfer limit that it has asked for, and also gives incentives to corporations to book the data load in advance.

My 2 cents!

​Mrinal|LinkedIn

Like others pointed out Dropbox is not using AWS S3 anymore, why? Well, they have come up with their own storage infrastructure called the Magic Pocket. MagicPockets are actually super servers that DropBox has created from scratch!

In an interview, the storage lead at Dropbox pointed out that any kind of written work since the time writing was invented, the documents storage size comes down to 50PB including all the languages, DropBox claims that their MagicPocket can store upto 500+ petabytes of data.

FYI, a single petabyte is a MILLION gigabytes, so yeah go figure!

But then DropBox started their journey with Amazon S3, so it has its importance. Why did they start with S3? Because of its affordable rates, seamless integration and the awesome scaling option!

To know more about AWS S3 you can check this blog out!