Google Cloud Alters the Deal

The first time I gave my “Myth of MultiCloud” talk, I pointed out that a common fear used by cloud doubters (one of which I used to be!) was “what happens when your cloud provider slaps a zero onto the end of the price tag?” My rejoinder to that was “we have over a decade of history of them not doing that exact thing; it’s fine.”

The next week Google promptly raised its Maps API pricing by up to 14x. In future versions of that talk I had to put a Google-shaped asterisk next to that point, but okay; it happens. Cloud is hard. Pricing is harder. Naming services well is apparently impossible, but that’s out of scope for today’s post. Every company gets a mulligan now and then.

And now Google Cloud has decided to once again increase pricing, this time for certain types of load balancing, data transfer, and storage. And I’m starting to think they aren’t exactly here for the hunting, to reference a particularly ribald joke.

What happened this time?

Let’s start by examining their blog post. Does it say it’s a price increase? Not exactly. The actual changes are convoluted enough that to effectively understand their impact I (a Cloud Economist) would have to go to some effort to model them out. I can therefore save significant time by saying it’s obviously a price increase, because if it were a net price decrease then they wouldn’t put that data inside of a confusing page linked from a blog post, they’d scream about it in the headline.

The various ways around the increase (special contracts with Google, committed spend locked in before the pricing changes take effect) are cold comfort. As an industry, we’ve seen a decade and a half of “anecdata” about a fundamental tenet of the cloud adoption tsunami sweeping the global economy. Specifically, that if you build something in the cloud it will over time become more stable, more capable, and if the price changes it will be a reduction instead of an increase. We now have clear evidence that that tenet is in fact untrue in Google’s case and not to be depended upon.

Why would a customer care?

If I launch a paid tier of Twitter For Pets and decide that I’m going to charge each user $9.99 a month, that price isn’t generally something I’ve arbitrarily picked out of thin air like I did the central conceit behind the ridiculous example company. It bakes in things like “Apple’s cut of subscriptions on the app store,” my overhead cost to run the company, and how much it costs me to serve my customers over the course of a month.

While many companies don’t have a great grasp on their unit economics (and I assure you, I’ve talked to a bunch of them, those hazy points are generally around not being sure how to allocate the cost of shared resources. They are not around “we are losing money on every transaction.” But when your cloud provider changes pricing dimensions and / or raises prices, that’s exactly what happens.

I’ve often said that cost and architecture are the same thing when it comes to cloud, and this is a particularly poignant example of that tenet. Customers design architectures and by extension products based in no small part upon how they have to pay for the underlying service usage themselves to deliver those products. If a customer designed pricing around accessing a Google cloud storage bucket at no cost, and now these changes mean that there’s a per-GB cost once they’re in effect, then there’s a potential for customers to have built a thing that’s suddenly no longer viable for them to sell.

What does this mean for customers now?

The most galling part here is that Google’s general responses around criticism of changes like this is how small of a change it is for the overwhelming number of customers, they’ll work with existing customers to find solutions, etc. That doesn’t solve the actual problem that Google just enthusiastically inflicted upon itself.

Those rejoinders don’t account at all for the frankly STUPENDOUS damage that’s been inflicted upon their future prospects, and it all starts when one founder or CIO turns to another over drinks and says “hey, what cloud should I use?”

“The data shows that most users” type Google response isn’t going to work here because the conversations that this manifests within are ones you’re not invited to, Google. These are where the real decisions are made.

As one of those people who gets asked questions like this from time to time, my answer now has to account for “well, today’s pricing is great but Google has demonstrated a pattern of changing the deal down the road in a way that AWS never has.” The great things that Google has going for it, such as its ease of use or its thoughtful opinionated design of many of its services all pale in comparison to “…but you might not have a viable business after their next pricing revision.”

This isn’t just a problem for small companies, either; the thing that large enterprises crave in their platforms is stability going hand in glove with predictability. This recent change resounds of Google simply Not Getting It, and that so thoroughly undercuts the very delicate messaging they’ve been putting forward with great care up until this point that I almost wonder if it’s the precursor to a pivot away from serving the enterprise cloud market entirely.

I have no insight into the internal economics of how Google Cloud makes its money, but I really can’t fathom that these pricing changes are significant enough to justify setting fire to its reputation the way that it has. I’ve been looking for the aspect that I’ve been missing, and I’m just not seeing it no matter where I look. Instead I’m seeing customers of all shapes and sizes looking at this change with confusion and horror; Google’s usual defenders are nowhere to be found.

I suppose this ushers in an era I was hoping to avoid: when describing the cloud provider choices, Google’s new description is invariably going to be “Google Cloud is great, but…”