Snap (SNAP -1.60%) may have finished off 2017 on a strong note, but the Snapchat parent isn't out of the woods quite yet. The company remains in the midst of a crisis regarding its redesigned interface, all while trying to scale its ad business and keep costs in check. Snap has made undeniable progress on some of these fronts, but investors should keep in mind that Snap has made substantial spending commitments to its two primary cloud infrastructure providers: Alphabet (GOOG 0.20%) (GOOGL 0.14%) subsidiary Google and Amazon (AMZN -0.13%).

The company was able to meet those spending commitments last year, but just barely.

Snapchat logo

Image source: Snap.

Commitment issues

These spending commitments ramp aggressively in the years ahead. Each agreement is structured differently. Snap has committed to spending $400 million per year on Google Cloud, its primary hosting provider, with escalating spending commitments at Amazon Web Services (AWS). Here's what Snap is on the hook for:

Year

Google Cloud

AWS

Total

2017

$400 million

$50 million

$450 million

2018

$400 million

$125 million

$525 million

2019

$400 million

$200 million

$600 million

2020

$400 million

$275 million

$675 million

2021

$400 million

$350 million

$750 million

Total

$2 billion

$1 billion

$3 billion

Data source: SEC filings.

And here's how much Snap spent on total hosting costs throughout 2017:

Quarter

Hosting Cost

Q1

$99 million

Q2

$106 million

Q3

$121 million

Q4

$131 million

Total

$457 million

Data source: Snap.

The agreement with Google allows Snap to shift up to 15% of its commitment to a subsequent year for each of the first four years. Indeed, it appears that the spending in excess of the commitment has been applied to subsequent years. In its 10-K, Snap notes its total hosting commitment for 2018 is $518.8 million, about $7 million less than the $525 million that is contractually obligated.

However, what happens is next is arguably more important for investors.

"If we fail to meet the minimum purchase commitment during any year, we are required to pay the difference."

The Snapchat design overhaul is in the process of rolling out to all users this quarter, so it's clearly still premature to call it an overall success or failure, even if user criticisms are mounting.

Since hosting costs are variable costs and are correlated to usage, investors should consider a worst-case scenario to appreciate the risks associated with these commitments. If usage ends up plummeting as a result of the redesign, Snap is still on the hook for that spending, and would theoretically be paying for basically nothing if usage declines. "If we fail to meet the minimum purchase commitment during any year, we are required to pay the difference," Snap writes in its 10-K. In the case of AWS, if Snap has to pay the difference, it can be applied as a credit toward future AWS purchases but won't count toward meeting future minimum purchase commitments.

As mentioned above, the company does have a little bit of flexibility -- $60 million worth for 2017 to 2020 -- so it could probably handle a modest and temporary decline in usage. But a significant and sustained drop could be catastrophic for Snap's balance sheet.