Snap's (NYSE:SNAP) cloud infrastructure strategy is unique for a company its size. The company's Snapchat platform relies entirely on third-party vendors for hosting. In this segment from Industry Focus: Tech, Motley Fool analyst Dylan Lewis and senior tech specialist Evan Niu, CFA, talk about the pros and cons of such a strategy.

A full transcript follows the video.

This video was recorded on May 12, 2017.

Dylan Lewis: You also think the company's hosting strategy as well is short-sighted, as well. This is something that a lot of companies haven't done. Rather than investing in their own infrastructure, whether it's building out their own data centers, or doing co-located facilities for their servers, Snap is relying entirely on third-party cloud infrastructure providers. That's kind of new. I know you've been a skeptic of that in the past.

Evan Niu: Yeah, it's a very unique strategy. If you look at it from Snap's perspective, the benefits they're saying of why they do it this way is, I think one is reliability and resilience, because AWS and Google Cloud are extremely powerful. That way, they can ensure the services are up-and-running. If you remember Twitter in the early days, it had outages all the time because they had so much trouble scaling because people were coming to the service so quickly, it was overwhelming Twitter's own servers, and their in-house infrastructure that caused constant outages. So, by relying on AWS and Google Cloud, Snap can avoid that, because those clouds are so strong that it's really unlikely they're going to be hit with an outage. Of course, it could happen. But there's much less risk of that happening. The other piece of it is, they're trying to save all this money on capital expenditures. Capital expenditures last quarter were very low. They're always low for Snap, because that's just how they approach their cloud stuff. It was $18 million or something. 

On the call, they mentioned, "We estimate that using a capital-light approach has saved us billions of dollars in capital expenditures," which is true, but if you turn around, they've also committed to spending $3 billion at AWS and Google Cloud over the next five years, combined. You're not really saving that money, you're still spending it, you're just, instead of spending money on capital expenditures and having that asset on your balance sheet where you depreciate over time and it's your infrastructure, now you're relying completely on these third parties. Those are variable costs. That's going to be a really big challenge for scaling, because those costs are going to scale, too. They did renegotiate their contract pricing during the quarter with both AWS and Google Cloud. They did say that lower contract pricing did help their cost structure a little bit.

And if you're looking at it from Amazon or Google's perspective, Snap is a huge customer in the space. They're probably one of the biggest customers out there with these types of spending commitments. These are the two biggest cloud providers, so of course they're going to be bidding pretty aggressively and competing for this contract, and I'm sure Snap is pitting them against each other and really leveraging that in negotiating. So, yeah, they are getting some better pricing now. But, it just doesn't seem like a good long-term strategy. If you really want to have good long-term growth, you just build infrastructure yourself, eat the costs up front, but then you can scale so much better.

Lewis: Yeah. You talked about it being variable and somewhat subjective to pricing negotiations. This most recent quarter, hosting costs ate up 70% of revenue, and that's actually down from the last couple quarters where it was up near 80% a few quarters ago. 

Niu: A year ago, it was like 160% -- hosting costs exceeded revenue a year ago.

Lewis: Yeah, I'm not sure how much further that's going to fall. We might continue to see it move down into the high 60s, or something like that. But, I don't know how much lower that floor is going to be. But, it really speaks to how their cost structure is very different than a Facebook, a company that has decided to build out, what, seven data centers at this point?

Niu: Nine.

Lewis: Nine? Wow. So, rather than have that fixed cost, like you said, as they have more people using the platform, and more people using the platform longer, the cost for them doing that is going to continue to go up, which is something that could be good or bad, it depends on your outlook and what you value for the business.

Niu: It gives them more flexibility if things start heading down, if people stop using the platform and users start abandoning Snapchat. Then it's better, in that situation, because you're not stuck with all this infrastructure you have built, and your costs will scale down. But, obviously, Snap is hoping things go up, in which case, it's really hard. They're getting squeezed by these costs, and I think it's really short-sighted. But to be fair, they have hinted that they might pursue their own infrastructure at some point in the future. But, they were very vague about it in the prospectus. They didn't specify timing, and, of course, they do have spending commitments for five-plus years. So it doesn't seem like if they do it on their own, it would be within the next five years, at least.

Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool's board of directors. Dylan Lewis owns shares of GOOGL and FB. Evan Niu, CFA owns shares of FB. Evan Niu, CFA has the following options: long January 2019 $20 puts on Snap Inc. and long January 2018 $120 calls on FB. The Motley Fool owns shares of and recommends GOOG, GOOGL, AMZN, FB, and TWTR. The Motley Fool has a disclosure policy.