Leading off an article on The Verge about the increasing rate at which photo-sharing startups are flaming out, we have this bit about Picturelife CEO Nate Westheimer, who had previously wooed customers of the soon-to-be defunct Everpix by touting his own company's rosy prospects:
But just a few months later, Westheimer acknowledged the difficult economics of storage-based businesses. Like Everpix, Picturelife’s millions of photos were stored on Amazon Web Services servers — and its costs were growing quickly. "Picturelife, a relatively nascent company, will pay Amazon over $1 million just this year," Westheimer wrote. "The way I see it, it’s a million dollars standing in the way of thriving and sending them even more money in the future."
Westheimer’s post was prescient. On Thursday, he sold Picturelife to digital media hub Streamnation for an undisclosed sum. All but two of the company’s eight employees, Westheimer included, will leave the company. And while Picturelife will continue to operate, its failure to remain independent is telling: it may be simply impossible to build a standalone billion-dollar business based around storing people’s photos. "This isn’t a product like Snapchat; it doesn’t take off overnight," Westheimer told me. "So you need that financial durability in the early years, while you’re still building your customer base."
The article goes on at some length about the paradox that while the photo-sharing business seems ripe for innovation due to the generally lackluster offerings of the major players, no one else seems to be able to make a go of it. The author never really comes to any sort of conclusion.
The clue is contained, I think, in the excerpt above, as well as in this paragraph that comes toward the end:
Perhaps more importantly, StreamNation is storing content on its own servers: a more difficult proposition than AWS, but apparently a much cheaper one. "Today Picturelife on Amazon is a huge monthly loss," Benassaya says. "Today PIcturelife, on our platform, is generating margin. And this is how you transform, just by making the right technological choice from the beginning." Picturelife will be six times cheaper to host on StreamNation’s own servers than it was on AWS, Benassaya says.
I am a big fan of Amazon Web Services. They encourage the use of architecture and infrastructure design patterns that can be highly scalable and resilient, and they allow companies to get off the ground without first spending a ton of cash building out a data center.
In the case of these photo-sharing apps, though, building all your stuff on AWS is basically a recipe for financial ruin.
S3 storage is remarkably cheap as storage goes, but it's not free. More importantly, because you are paying for a service (not purchasing physical assets like servers and storage arrays), you can't capitalize those costs over the lifespan of the asset.
More importantly still, the big cost advantage of using Amazon's services is that because you can rapidly deploy and tear down infrastructure, you don't have to build out all the stuff you will need to support peak usage of your product and then let it sit there idling at 2% utilization the rest of the time. For a photo-sharing service, I would imagine there is a bit of oscillation at the compute tier around holidays and vacation times (i.e., the times when people are most likely to be uploading and sharing photos), but mostly what you have is an ever-growing storage tier.
In other words, your hosting costs are just going to keep going up, and your accountants don't get to spread those costs out over the span of three to five years. They hit your bottom line every time the AWS bill arrives in the mail.
Is it any wonder that companies for whom this is their primary business keep going broke?