Amazon was not profitable in 2022. On about $514 billion in revenue for 2022, Amazon generated a net loss of 2.7 billion. It was since 2014 Amazon didn’t record a net loss. Indeed, in 2014 Amazon reported a net loss of $241 million, and it was profitable until 2021. In 2022 Amazon turned unprofitable again.
Amazon aggressive growth
While it managed to survive the dot-com bubble Amazon also hat to tweak its business playbook throughout those years:
Amazon’s business model today has several moving parts:
- Amazon e-commerce.
- Amazon platform business (Amazon has hosted third-party sellers on the platform since the early 2000s).
- Amazon AWS.
- Amazon Prime.
- And more.
Below you can appreciate the Amazon business model in full:
As of 2021, Amazon AWS is a major contributor to Amazon’s operating and net income.
That is because AWS has different unit economics logic:
Born as an infrastructure built on Amazon’s infrastructure, AWS has become a company offering cloud services to thousands of clients from the enterprise level to startups.
And its marketplace enables companies to connect to other service providers to build integrated solutions for their organizations.
Amazon’s cost structure today
The interesting take about Amazon, which many are not aware of, is the fact that its core business (the e-commerce platform) is also, and still, unprofitable.
As you can see, both the North American and International segments are unprofitable.
Nonetheless, in that segment, there are businesses like Amazon Ads and Prime, which might be running at excellent gross margins!
And the picture is even more impressive if we look at Amazon’s profitability without AWS!
It’s easy here to dismiss Amazon and say, “wow, after decades in business, the company is not yet profitable.”
But wait for a second.
Before you dismiss it, there are a few considerations to make.
Bits and atoms
First, Amazon isn’t just an e-commerce company. Amazon is about inventory and fulfillment as much as it’s about e-commerce.
That’s the nature of its flywheel!
Inventory and fulfillment are very intensive in terms of capital requirements, and yet they are critical for enabling customer experience.
And the good news? They give the company much stronger moats.
Over time, anyone might be able to replicate Amazon’s e-commerce.
But a combination of e-commerce, inventory, and last-mile delivery?
Extremely hard to replicate!
Last-mile and transferable network effects
But when they do kick-off, they might make a tech company valuable for years.
There is another critical point about it: transferable network effects.
Building, maintaining, and speeding up network effects is extremely hard.
Do you know what’s harder? Building liquid network effects.
In short, a liquid network effect is when a platform business has built such an infrastructure that there is plenty of supply and demand to rely on, and those sustain themselves in a sort of smart dynamic market.
When that happens, a company that has empowered such liquid networks can go on and try to transfer them across a new industry.
One example of this is how Uber, starting from ride-sharing, first expanded this market to become a multi-billion dollar one.
Then it managed to expand into adjacent segments like delivery and freight.
When you have transferable network effects, you don’t think about a single industry but start thinking about the entire industry, which can be unified under a single paradigm.
For instance, you don’t want to call Uber a ride-sharing, delivery, or freight company today.
Instead, you want to call it a Last-Mile Platform!
A last-mile platform can tackle any industry which relies on the last-mile problem, which states that the most challenging part of a delivery network is in the last mile (or, if you wish, in the last steps) from the company to the customer.
These last steps in the networks, indeed, are a trillion-dollar issue.
These last steps fall outside the network, making it fragmented, unreliable, and expensive.
Thus, when you re-frame the kind of problem a company like Uber and perhaps Amazon is trying to solve, you understand the real potential value of the network!
In short, the sort of network effects that might make you able to launch a whole new business much more quickly by simply leveraging on the existing tech platform!
While figuring out how to unify the networks and make them less and less fragmented.
If you can figure out that problem, you can transfer it across many industries, thus, redefine them!
Built for scale!
When we look at Amazon’s e-commerce platform, it’s critical to consider that since the onset, it has been built for scale and reach.
In short, the e-commerce platform aims to enable as many customers as possible via convenience, variety, and service.
Amazon might make money from it in the future, but it might well be that in 20 years, the e-commerce platform will still be primarily run for scale and reach through convenience, variety, and service.
And it’s worth remembering that thanks to this strategy, Amazon was propelled into “Walmart Status!”
It took Walmart sixty years to get there, while Amazon took less than thirty years…
The digital ads empire
Now take the case of the Amazon e-commerce platform and Amazon Ads.
And to put things in context, Amazon Ads were larger than YouTube Ads in 2022.
With a core difference, Amazon’s ads segment is just one of the many moving parts for the company!
And we can easily guess that the ads segment might be highly profitable and scalable.
So, if Amazon scaled this to a hundred billion per year business, would e-commerce finally become profitable due to the ads platform?
The AWS rocket ship
In the meantime, Amazon AWS keeps growing at a staggering rate.
Of course, as competition in the cloud industry intensifies, this might slow down revenue growth and profitability.
However, it’s worth pointing out that now only is Amazon AWS a tech giant for its own sake, but also how instrumental it will be for the current AI revolution.
Indeed, AI models, that rely on massive computational power through AI supercomputers need an infrasctructure like AWS to run in the first place.
Take the case of how Stability AI has pre-trained Stable Diffusion on top of AWS!