Dropbox's Best Move Was Breaking Up With Amazon's Cloud
Dropbox's Best Move Was Breaking Up With Amazon's Cloud
(Bloomberg Gadfly) -- Four paragraphs in Dropbox's IPO filing highlight the limits of cloud computing for at least a subset of companies, with ramifications for Amazon, Google and others.
Those paragraphs in the public offering document (page 67) summarize the difficult and nerdy work to shift a vast volume of Dropbox users' digital files from Amazon's computer networks to Dropbox's own and to close dormant accounts to free up storage capacity. This yearslong shift to wean Dropbox off Amazon Web Services wasn't glamorous work, but it improved Dropbox's finances substantially. Without exaggeration, the shift away from cloud computing is one of the biggest reasons Dropbox is able to go public now.
What Dropbox did, however, isn't necessarily a great selling point for the Amazon.com Inc. cloud-computing service or those of rivals such as Google and Microsoft.
Like many young technology companies, Dropbox started life relying on AWS. It had its own data centers and computing equipment, too, but paying to outsource much of its computer infrastructure to AWS allowed Dropbox to avoid much of the cost and distractions of buying flotillas of equipment, erecting buildings to house it and paying for electricity and people to run its computing backbone.
At some point, however, AWS was no longer so good for Dropbox. The company decided a few years ago that it was more cost effective, efficient and strategically important to build its own computing network from scratch and tailor it to its needs. Dropbox still uses AWS, but the vast majority of its user data is stowed on its own infrastructure.
That decision helped transform Dropbox's finances. In 2015, Dropbox's gross margin -- or the amount of its revenue left over after paying its basic costs such as storing and distributing its users' files -- was 33 cents out of each dollar of sales. That's pathetic for a software company. Box, Salesforce and HubSpot have gross margins of 70 cents or more for each dollar of revenue. (Although Dropbox is a different animal from those business software firms.)
Dropbox was also bleeding cash; part of that was due to spending gobs of money in 2015 and 2016 to build the bedrock technology to pull away from AWS. But even excluding that spending, it was clear Dropbox was having a hard time making its numbers add up when it was paying outside companies for much of its computing back end.
Moving to its own infrastructure drastically improved cash flow and gross margins, which were 70 cents on the dollar in the fourth quarter. Dropbox also persuaded more of its users to pay for its software. Without these financial improvements, I doubt Dropbox would be a viable IPO candidate today.
The broader question is what Dropbox's shift away from outsourced cloud computing says about the future of this technology category. Will more companies decide they have outgrown the cloud?
I doubt it. Most companies are not in Dropbox's situation. The company's core business is storing files efficiently online and retrieving them seamlessly. That means it derived huge benefits from a customized, in-house version of AWS, and few other companies could. Dropbox also has the skills to operate its own computing infrastructure, and again few other companies do.
It is interesting, however, that two of the biggest tech IPOs in the last year made the opposite choices about computing infrastructure. Snapchat's parent company relies almost entirely on Google and Amazon cloud services to zip its users' photos, chat messages and videos around the globe. Eventually, Snapchat's cloud-computing reliance could make it wildly profitable because it doesn't have to pay to build and maintain its own technology infrastructure. For now, though, the company's cloud-computing spending is a reason Snap Inc. is burning cash at a level I've never seen before and had negative gross margins for many quarters of its life.
It's clear there's no right answer about whether it's better to own versus rent the technology backbone for computing activity. Netflix famously leans heavily on AWS, although it also operates its own technology to ensure its streaming videos zip faster to customers' homes, and it has sophisticated homegrown technology to make AWS work best for Netflix's needs. I talked last year to Indian e-commerce company Snapdeal, which shifted off cloud computing to its own custom-built infrastructure using an open source technology called OpenStack. A company executive said the shift was painful but shaved Snapdeal’s infrastructure bill by 75 percent.
In other words, the cloud isn't for everyone. That's not necessarily a message you'll read in Amazon or Google marketing materials, but it is the reality. And it is clear and ironic that Dropbox -- which sells a cloud-computing product -- found salvation in mostly ditching the cloud.
This column does not necessarily reflect the opinion of Bloomberg LP and its owners.
Shira Ovide is a Bloomberg Gadfly columnist covering technology. She previously was a reporter for the Wall Street Journal.
To contact the editor responsible for this story: Daniel Niemi at dniemi1@bloomberg.net.
©2018 Bloomberg L.P.