Investing (at a loss) many years to grow a business where you are starting behind others seems sensible. Let's not forget that Amazon had reported almost no profit or just loss for more than a decade. So as long as the business grows it seems like things are moving in the right direction.

Only if you think you can get the costs down.

It's software as a service. Of course you can.

> It's software as a service. Of course you can.

Which is directly tied to the underlying hardware. That they already optimize to death.

All cloud providers charge several times what the underlying hardware/electricity/peering is actually costing then. If the goal is to become immediately profitable rather than continue growing, then just stop reinvesting money into the business in the form of hiring and new data center construction.

Have you seen how much other providers charge? It's the software you're paying for when you go with google, hardware is cheap.

It's also hardware as a service... and, crucially, customer service. Google is awful at that.

So if I sell $1 bills for $0.90, I guess I should be glad if my “revenues grow”?

Do you have any evidence to believe Google Cloud Platform is negative gross profit?

My background in the industry would lead me to believe GCP (just the infrastructure, not the applications) is 70%+ gross margin with the hardware (and data center) cost (depreciation if capitalized) above the gross margin line.

For a business at that growth rate running a GAAP negative net income is easy to do while still building a high quality profitable long term business.

I’ve made no statements either way. I was responding to the relatively naive idea that increasing revenue without profit was a measure of “success”.

But, cloud companies don’t just have to worry about hardware and infrastructure. They also have to worry about software engineers, “enterprise sales solution architects” and a whole host of other costs.

In logic class we called that a red herring fallacy. Google Cloud isn't in the business of selling $1 bills for 90 cents.

The “fallacy” is offering evidence of success by just bringing up revenue. I made no claims on whether GCP is profitable or not.

But we do know AWS's operating margin which is 29%. Spending a little to gain market share in a high margin business seems sound to me.

If you expect to be able to acquire $1 bills for $0.80 in the future, then yes.

Because Google has historically been a shining example of a company that could create profitable lines of business outside of advertising....

There's a difference between selling something with a fixed cost (and a fixed value) for a fixed price and selling something which has unknown (and variable) value for a fixed price.

So there are no variable costs in servers?