I don't think Amazon Advertising should be considered separately from Retail.
A huge portion of it's money comes from people paying for the top search result on Amazon's retail site. Oh sure, they ad a very tiny "sponsored" but you have to scroll past three of those before you see the actual search result.
But if Amazon Retail didn't exist, Amazon Ads wouldn't either. Meanwhile, AWS could be split off as an independent company and be quite successful.
Advertising used to be treated as a Retail line item, but Bezos deliberately broke it out to put profitability pressure on Retail by not giving them a way to hide their losses in the bottom line.
We agree, AWS is a clean break, ads is way more messy.
But ads only works because of Amazon retail's sorta-kinda-monopoly position. It's still a big distortion of the market. But I don't have a bright idea what to do about it.
Isn't it kind of self-correcting? The more Amazon pushes products sold by third parties, the worse shopping at Amazon becomes. The worse shopping at Amazon becomes, the more people will shop elsewhere and not see the ads.
Absolutely. Every HN thread about Amazon retail is littered with comments about giving up on shopping there and finding a better experience (e.g., less junk, cheaper prices, better selection) elsewhere. If anything, the growth of Amazon Ads has pushed e-commerce spend to Walmart, the actual leader in retail.
I agree 100%. However, a lot of my woes with Amazon are due to inventory commingling- because I make sure to only buy Ships & Sold by Amazon items.
With that said, Walmart is beginning to push 3rd party sellers a lot online too, and filtering for items sold by Walmart can be difficult. A lot of times items show up as in stock in search results only to actually not be in stock when I finally load the product or cart pages, and Walmart will often switch these to in-store pick up automatically which is frustrating as the nearest store to me is quite far. I’ve had good luck shopping at Target but their prices are almost never as competitive.
That sounds like a frustrating experience. I don't shop at Walmart because Seattle is kind of a Walmart desert (whereas Amazon return drops are everywhere), but it always seems like there are comments promoting it in those threads.
lmao as someone who spent a couple months at Walmart global tech, they may be the leaders in retail but they are not stealing Amazon customers. If anything Target, Wayfair, Best Buy, TJ Max etc are taking more market share.
Last time I shopped at Walmart, it was a mess and their e commerce is even worse, products come all beat up, 3rd party policies are a free for all atm, and the scams are even worse there
The other stores' taking e-commerce share would be consistent with the spirit of the comment, IMO.
I agree on the third-party products/listings. I have to assume that Sam Walton is turning in is grave because it all seems like a perversion of old-school merchandising.
Amazon retail is a tent pole... Not unlike a blockbuster movie that might not even be profitable, but creates profits for all kinds of businesses around it.
I'm not sure we do. Amazon retail at this point from a business perspective is nothing but a way to sell ads. Talking about retail losing money when not counting ads is like saying google search loses money when not counting ads. And no, Amazon would still be able to sell ads if they had a smaller market share.
I was going to make the same point. Their retail business could lose $billions if people are paying more $billions to advertise on top of the same platform.
It’s a smart move to use single digit margins in retail as a route to making double digit margins in online advertising.
+1. Not much different than Google probably losing a ton of money on core search to create a market for what does make them money (selling advertising on search).
Per SEC rules, once a business reaches a certain size it has to be detailed out on the financials. That's why the ads business is reported separately now
> "In May 2014, the Financial Accounting Standards Board (“FASB”) issued an Accounting Standards Update (“ASU”) amending revenue recognition guidance and requiring more detailed disclosures to enable users of financial statements to understand the nature, amount, timing, and uncertainty of revenue and cash flows arising from contracts with customers. We adopted this ASU on January 1, 2018 for all revenue contracts with our customers using the modified retrospective approach and increased retained earnings by approximately $650 million."
The only section in the 10K referencing FASB is above.
This is related to reporting CUSTOMER contracts, not a requirement to report Business Division revenue (like Amazons Ad Business).
The reason for this change, which again is unrelated to the original topic at hand, is that it posses a financial risk to investors if X customer represents a significant portion of the companies revenue.
> Certain advertising services are now classified as revenue rather than a reduction in cost of sales
Reading in-between the lines on this, advertising flipped from retail contra-cogs to it's own revenue segment. Advertising biz got to big to report like that.
> A huge portion of it's money comes from people paying for the top search result on Amazon's retail site.
What are some of the average metrics for the average Amazon seller doing paid advertising? Cost per impression, cost per click, cost per conversion/sale?
T-mobile and Sprint has been a win for consumers - as much as I wish we had dozens of carriers instead of 3. Sprint was a zombie corporation and T-mobile was well run but lacked the spectrum to compete. One of the few times merger logic worked out even better than the ibankers promoted.
Microsoft has embedded itself deeply into the democratic party through huge PAC donations over decades. If anyone has sway with the democratic appointed FTC, it would be them.
Acquiring AWS would absolutely be the kind of emergency/pull out every stop to make it happen event worth tapping every connection (and then some) to make happen. Overnight they would go from scrappy second place to the dominant provider of all cloud computing on the planet.
Far more dumb mergers were greenlit in the past like AOL/Time Warner.
In the 2020 presidential election year Microsoft donated over 21 million dollars, almost all to democrats and the Biden campaign. They were the 26th largest donor, period.
Biden appointed Lina Khan, she owes her job to him (and he can choose not to appoint her again!). Biden owes a substantial amount of his presidential campaign funding to Microsoft... not hard to connect the dots here.
Go back decades--Hillary Clinton, Obama, Gore, etc. and Microsoft was always been donating enormous amounts of money to democratic candidates.
Again, you should really read up on what a PAC is before you form your conspiracy theories. Those are donations from Microsoft employees - not from the company, hence the big bold warning label:
"NOTE: Organizations themselves cannot contribute to candidates and party committees."
Wow, the 26th largest donor, that sounds like a threat to democracy. Microsoft also gives money to republicans.
How about an actual threat, the Koch family (one of the infamous brothers died), who by themselves have a budget about the same size as the entire us republican party, and have taken it upon themselves to select the next republican presidential candidate - and they will back that up with enormous spending. Regardless of your views on the political parties specific choices, it's a bad idea for a private group to control (or credibly attempt to control) who can run for president - this is fundamentally different than microsoft's donations (even if I don't like that corporations are putting money into politics). Koch have had a lot of success in setting the previous few decade's political framework about how the us looked at international trade, along with influencing the libertarian party (David Koch was the Libertarian VP candidate in 1980, but had enormous impact on that movement over time).
Google's market cap is $1.33T. Amazon's market cap is $1.04T, of which most (possibly more than all considering that retail is operating at a loss) is the value of AWS. Buyouts of public companies routinely come in at 20-30% above the current stock price.
Tim asks about why the $-19bn free cash flow for '22 is not a worry.
All of this is because enormous levels of cap-ex spend: $58bn in net purchases of property and equipment over the year ($63bn gross). The other piece is $9bn in debt paydown of lease obligations, which is not really concerning because it's just a debt reduction.
They don't provide much of a breakdown of this very large capex number. Presumably a lot of it is fulfilment center spend, but there's data centers, vehicles, etc., to consider as well. It really comes down to whether this spend is wise. If this cap-ex will generate strong ROI, it's fine.
Everyone seems to be talking about the YoY percentage growth decline for AWS being at "only" 20%. Seemingly what's happening is just optimization en masse which is temporarily impacting growth on a YoY basis. There are other comments that allude to the on-prem to public cloud migration being in its infancy and I agree with that despite the growth rate decline becoming more noticeable.
Everyone is signing up for Savings Plans and RIs. Everyone. We actually just released a report on this showing a steep decline in on-demand utilization throughout Q4 which you can see here...it's just amazing to see it finally popping up in earnings reports: https://www.vantage.sh/cloud-cost-report/2022-q4
The data on on-demand percentage change is really interesting. I wonder if there's some underlying correlation that causes people to under-estimate this transition. For instance, orgs using reserved instances might tend to be larger (lower relative usage variance) and therefore have fewer cloud employees per compute spend. Thus, the on-demand share weighted by employees would actually be far larger than when weighted by compute spend (and making the transition less tangible to the hivemind).
Thanks for publishing these reports! They have tons of business insight and, personally, I've really enjoyed reading them.
TLDR: If you have predictable money coming on because you are so necessary to society (like a utility), instead of paying profits to the government, you can take loans against your cash flow and grow by acquiring businesses with the extra cash. You will always seem unprofitable, and you get to benefit from loan interest deductions on your predictable cash, but in reality you grow bigger than ever, through acquisition. Another thing Amazon has going for it is the 30-60-90 terms it has with many suppliers and affiliates - they make the money now and pay them a month or three months from now. That makes them further bank in huge valuation compared to the rest of the sector. Cash flow is king.
That is why Leveraged Buyouts were a big thing for a while and why when interest rates go up, the math for many of these levered giants stops working or needs adjustment. Toys RUs are an example. It’s a very valid growth approach, and it works well if you can afford (and don’t overdo it).
Yes, I do, because as a software engineer my whole life I wanted to know how to actually capture the gains my code generates. Best way to recognize patterns and learn: read the annual statements of public companies in different industries in detail. You’ll notice quite interesting things - such as web hosts only growing 7% per year, or especially with earnings seasons moving markets now - seeing which companies are dependent on debt and worse, impacted by the higher interest rates, and which ones are growing like crazy despite that.
you should go out and look at whole food shops, walk in the streets early at delivery time to see in the trucks what kind of package people are getting (PRIME is now a word of english vocabulary), what video people have on their door, how many people hire amazon, how they have airline, all it takes is noticing all those daily influence that amazon has on professional and individuals life. they are the biggest company that has ever been in the western world. by a huge margin. and big is scary to government. big is scary to people. better look big and broke than big and financially looking like you killing everyone else
How much of Amazon’s airlift is in-house vs Atlas, ATSG, and Kalitta? I mostly see and hear the latter three, but I admit to not knowing just how tightly controlled or owned they are vs just contracted lift.
If AWS is spun off, then I don't believe retail would be able to survive. Online retail vs. brick and mortar still has a lot of growth left and it won't make sense to spin it off until it grabs an even bigger piece of the pie.
The Amazon retail business would struggle to compete effectively on its own, and would have to work on its weaknesses. As a whole, the economy would be better off.
The essence of US antitrust is consumer harm. If Amazon were broken up the retail arm might die, and consumers would arguably suffer -- they wouldn't receive the subsidies they currently get from AWS, they'd pay more for products and get slower shipping.
To argue that Amazon retail consumers would be better off without the subsidies from AWS takes some slightly more complex logic -- that increased Amazon prices would lead to more competition and so lead to lower prices (or better quality). IMO that argument is tenuous.
If Amazon, having killed off its competition, were to raise prices I think you'd have an argument, but at the "undercutting" stage I think (IANAL) they're safe.
> If Amazon were broken up the retail arm might die, and consumers would arguably suffer -- they wouldn't receive the subsidies they currently get from AWS, they'd pay more for products and get slower shipping.
This isn't necessarily a downside for consumers as you might get actual customized retail shops opening up again. This would enable you to get your stuff faster than delivery.
I would love to have a "technical bookstore" back again, for example. Amazon's retail arm imploding would enable that business to return.
How far do you take this? An individual developer would struggle to compete effectively on its own, and would have to work on their weaknesses. As a whole, the economy would be better off. So should we break up every venture of more than one person?
Retail survived for over a decade before AWS launched (and when AWS launched, it obviously wasn't carrying the company for many years after that).
Why do people think Amazon retail couldn't survive now without AWS, when it survived just fine before AWS? (And now has Prime, advertising, and 3PS/marketplace fees as a logical part of the Retail/non-AWS unit.)
I believe that if "storefront" was spun off of AWS, AWS would falter less than a decade.
Having an anchor tenant with "2014" problems in 2010, or "2020" problems in 2015, especially if they roll up to the same CEO as your org, gives the org a huge head start in making software and platforms that will, once 2020 actually rolls around be tremendously compelling to everyone else.
Amazon provides AWS with problems to solve that are going to be general problems to solve in 2-3 years. Amazon developed DynamoDB and S3 and handed them off to AWS to operate, for instance.
Without Amazon's innovation funnel, AWS will be Rackspace in 10 years (well, maybe Oracle in 5 years). Maybe they'd keep adding more hosted services (Announcing ClickTower, our columnar database as a service!) but that's not innovation.
> If AWS is spun off, then I don't believe retail would be able to survive.
At a really high/overly-simplified level, how far off from probability is their retail section? Do they just need to semi-aggressively tweak some margins about 2-3% in maybe their shipping cost/subsidizing department, slow down on R&D/reinvestment into this sector by about 2-3%, and maybe swing their marketplace fees 1% and boom, they are in business?
Or is it more complicated?
I wonder how much fat there is to trim in this sector of their business, if any.
I wonder what we will see from AWS as we go through an economic cycle though.
Great business when everyone wants to scale up.
Maybe not a great business when people want to scale down.
> For contracts with original terms that exceed one year, those commitments not yet recognized were $110.4 billion as of December 31, 2022. The weighted average remaining life of our long-term contracts is 3.7 years. However, the amount and timing of revenue recognition is largely driven by customer usage, which can extend beyond the original contractual term.
Even if new business stops, and that revenue extends beyond contracted terms, it should provide a nice cushion.
AWS 20% Y-o-Y growth.... My take away from this is cloud compute utilization is still in its infancy and will continue to grow faster than most sectors of the tech economy. There's still a lot of stuff on-prem, but new deployments are mostly cloud and that trend will continue.
As people get more used to cloud-native deployments, they'll see the value in rented compute instead of leased CPUs. It just seems like too good of a bargain (even with the high prices of AWS) to use them instead of self hosting. You need less engineers, you get pre-built tech stacks and there is a major reduction in support burden.
> My take away from this is cloud compute utilization is still in its infancy and will continue to grow faster than most sectors of the tech economy. There's still a lot of stuff on-prem, but new deployments are mostly cloud and that trend will continue.
So what I’m curious about is whether the growth is actually in transitioning from onprem to cloud, or existing cloud consumers requiring more infrastructure.
I know we’ve been throwing more and more compute at our scalability problems rather than optimizing what we have.
I've wondered the same. Anecdotally every company I've worked for in recent years made the move to cloud years ago, but every year they require more infrastructure.
I'd actually be surprised if most of the growth in cloud was coming from companies migrating at this point. From my personal experience it would seem more likely that most of the growth would come from the growing demand in digital infrastructure, not demand for cloud per-say.
> I know we’ve been throwing more and more compute at our scalability problems rather than optimizing what we have.
On this point, I've noticed that cloud has made it so much easier to add infrastructure that companies will now do it without really thinking. For better or worse hardware is no longer the constraint it used to be.
I think the driver is new deployments. If you're starting from scratch, you'll probably start with the cloud. New companies are created all the time and old companies tech ages out. I'm guessing when faced with a new deployment, the three main cloud vendors will be the road taken.
Everything either starts or ages out at some time and is created.replaced with something new, and when that happens, its going to be in the cloud.
Tim argues that Amazon should spin off AWS, or perhaps he's arguing for government intervention to that effect.
What do we think would happen to the retail Amazon if it did not have the giant profit-center of AWS? Presumably it would crumble under a lack of profit, or start doing the MBA thing to generally worsen everything to drive up the quarterly results.
I don't understand the anti-trust argument against Amazon Retail and AWS. Amazon Retail went vertical by building their own cloud solution and then sold that solution to customers - what's anti-trust about this?
The folks who don't have a subsidized cloud solution have to eat some of their margin paying for services like AWS, which Amazon Retail gets effectively for free/others pay for.
I still don’t get why it’s a problem. There should be an advantage in taking the investment needed required to go vertical and leverage that for your entire business.
Depending on how much of the value goes along to AWS... retail could look like a lot of high value / low margin businesses that do just fine. Obviously if you give 99% of the value to retail and pull the profit center, it would be bad. But that wouldn't happen.
I don't quite understand the question. One wouldn't try to split the market cap. Instead you'd split the company (each share in Amazon becomes on share in Nu-Amazon and one share in AWS). The prices of those new shares will determine the market cap of each half.
You don't split the market cap of a company? Really?
If I have a company worth 1 billion dollars and split it exactly down the middle in assets and liabilities, the assigned value of each company is nowhere near 500 million dollars?
Yes you will lose a little bit in terms of "lost synergies", but the starting point for a split (or a merger) is to add together or divide up the market cap.
...and the resulting market caps will have a ratio. Brianwawok was wondering what the ratio would be.
If the multiples of each business were the same, we'd expect a roughly 75/25 split along the lines of profit, but the multiples are probably not the same, and that's the interesting part of the question. It represents market judgement on the future growth trajectories of e-commerce vs e-infrastructure.
I would like to read it that way, but then the GP's "if you give 99% of the value to retail and pull the profit center, it would be bad" comment makes no sense. It is suggesting the value is assigned, rather than found via price discovery, and the assigned values would not match the underlying assets.
Again, I would love to read it some other way, but there is none that makes sense. In your reading, how does one put "99% of the value" in one part but "the profit center" in another?
brianwawok's post uses the "give a value X to Y" colloquialism, which usually means "have an opinion that Y has a value of X." You seem to be reading it as "impose price X on Y through socialist decree," or something. I thought that was uncharitable.
Sorry, I have never heard anyone use "give a value" like that. I'll trust you that it's idiomatic, but even with that substitution it still doesn't make sense.
"Obviously if you have an opinion that retail has 99% of the market cap and pull the profit center, that would be bad".
What in the world is that supposed to mean? Our opinions aren't going to affect the world, there will no consequences if somebody has that opinion. How is it bad? And with that interpretation, I have no idea of what "pull the profit center" could mean. Because that has to be talking about the way you split the company, not about an opinion you'd have.
No you wouldn't. You'd have one share of AWS and one share of Amazon, that the market previously priced the sum of them as $1T. On market open, the market will have some volatility finding the estimated price difference between them, but I wouldn't expect any actual sales of Amazon retail at around $500b.
It depends on how would you split the corporate debt between the two entities.
If you load up enough debt on AWS, and have Retail debt free and with cash - it is possible to have both shares valued at the same price
Remember that Market cap = Enterprise value - Net debt.
We can argue that enterprise value of business is different, but by changing net debt figure of each company we can make market cap of both firms equal
I think the claim is that there are synergies to having both so the valuation of two pieces separately would be lower than together. How much would be this discount I am not sure.
Well, it wouldn't be the "MBA thing" so much as that, if you want to be a public company with shareholders (or even a private company really) you have to have some path to profitability. You can't just lose money forever.
Right the MBA thing is where you generally worsen things in search of quarterly improvement. "Yes we're making some money, but we could be making more money."
> I wish the author had elaborated on why it should be illegal. (Or maybe, how it is?)
Its implicit - an actor is taking over entire societal logistics and ecommerce while burning investor money with no obligation to make profit, hence be sustainable. (applies to many tech giants). Practically becoming a public utility that the society depends on for running itself day to day.
If that actor goes tits up, then the entire society will experience chaos due to that supposedly private, but actually public infrastructure crumbling down with nothing at that scale to back it up.
Its too risky and irresponsible. You can bet that they will force the government to bail it out if that happens, by saying that it is 'too big to fail'. Socialized costs. Privatized profits.
A huge portion of it's money comes from people paying for the top search result on Amazon's retail site. Oh sure, they ad a very tiny "sponsored" but you have to scroll past three of those before you see the actual search result.
But if Amazon Retail didn't exist, Amazon Ads wouldn't either. Meanwhile, AWS could be split off as an independent company and be quite successful.