[Pxtl]

I think you're missing his point. He's focusing purely on the cost that increment per-user - that is, the costs that are directly associated with each user.

The fixed costs are allowed to be obscenely high. Growth will overcome that. If you build an obscenely expensive server farm and spend $Xmillion developing software, you can get that back if you get X paying customers eventually.

However, if each user you get means you have to fork over another $12/mo to Amazon when the user only is paying you $10? There's no way to make that work. More users would actually cost you more.

Maybe there's way they could've torn out their infrastructure and rebuilt it as self-hosted. Maybe there were some optimizations they were missing that could've cut those cloud-based costs.

But on the surface? Every dollar the user handed them got handed right off to Amazon, and Amazon's prices go up as you get more users.

[barrkel]

Amazon is a way of bootstrapping quickly. You pay a premium for not managing the physical assets yourself. But I'm not saying anything you or I don't already know.

Of course they could have stayed in the black by growing more slowly and managing costs better. I think that's obvious, and uninteresting. It seems clear to me that the guys at Everpix were making a somewhat desperate effort to get VC traction. IMO that's what led to their increased burn rate.

I don't think they didn't know they were burning their reserves, that their cost structure was unsustainable.

I feel we may be talking past one another.

[ivanplenty]

I really appreciate yours and the GP's comments because both describe different parts of the Everpix puzzle.

You're right that capital should seek the highest returns, but one way to measure the likelihood of getting that return is by evaluating the marginal costs and revenues of a product. When a company sells each product for a loss, it is impossible for the company to provide a positive return on the capital. In those cases, like with Everpix, it becomes a question of "when" and not "if" the business will fail and the return will be zero. The only rational way to play the game that way is to hope for an acquisition.

That's why I look at detailed parts of business models like this, it helps elicit the overall picture in the same way functional a test case elicits overall product health. There is an art to ensuring proper overall coverage with multiple tests.

> I don't think they didn't know they were burning their reserves, that their cost structure was unsustainable.

This is where I disagree and why I wrote the analysis. I think the company didn't understand they were selling their product for a marginal loss:

> "Long story short, the infrastructure was paying for itself through subscription revenues." https://news.ycombinator.com/item?id=7041640

> "AWS infrastructures costs were already being covered by subscription income." https://news.ycombinator.com/item?id=6676906

From the numbers neither statement was true, and from the confidence in the tone it seems like they didn't know for a while. Subscriptions did not cover AWS costs. It looks like it might have become known internally when asked directly from others taking a look:

> "The reason we were getting closer and closer to being positive on variable costs ... is, yes, improved monetization, but more importantly AWS optimizations." https://news.ycombinator.com/item?id=7043555

[petervandijck]

You forget: the 10$/user to Amazon halves every 18 months. The $12 stays the same.