It is definitely the case that YouTube gets lots of really cheap compute by being part of Google. But that isn't a bad thing, they are making use of an otherwise wasted resource, effectively lowering the cost for products that need daytime capacity to serve humans.
https://blog.youtube/inside-youtube/new-era-video-infrastruc...
Vimeo is paid. None of the other big companies seem willing to subsidize an online video service to the tune of tens of billions per annum.
By nearly any standard, Google's control of YouTube should be a prime anti-trust target.
And how does “providing compute” even matter, YouTube is just using google servers like every other google service. Not like they’re getting something for free because they’re part of the same company.
You speak as if providing compute is their primary business. Youtube, existed long before their compute. They already had all their own compute. They only got into providing compute because they figured they could beat AWS with their own infrastructure.
I don't think anyone's in doubt YT is a monopoly with how big and widely-used it is. It's just that they continue to survive because of how well they execute their plan of making a good service that gives users what they want (fast, instant access to videos that appeal to them) and pays out creators enough to where they can make a living, or even build a media company (LTT), off of the adsense and monetization opportunities YT enables.
YT isn’t a monopoly however. It might be dominant depending of how you define the markets but it does have competitors.
That's just like saying that car manufacturers can't use idle capacity to help start a new line of vehicles just because a new car manufacturer couldn't create a new car line.
Here we have Google using its dominant position in another market to prop up another activity through an advantage. That’s most definitely fall under anti-competitive law at least in Europe (I don’t care about the US. The modern interpretation of American antitrust laws is a complete joke).
It's silly to say that companies should never be allowed to branch out to other businesses.
Care to highlight exactly how Netflix used its dominant position in film renting in favour of its streaming business?
It would indeed be silly to say companies can’t branch out. Thankfully no one is so that’s fine.
Apple had a dominant position in selling music and in music players by 2006. It was the number one music retailer in the US. It very much used that to get into the phone market and part of its early success in the App Store was that it was built on top of iTunes and it already had more credit cards on file than any other company in the US besides Amazon.
If you're not arguing that a company shouldn't branch out, are you arguing that once a company gets a certain size, it shouldn't be allowed to invest those profits and resources into another industry?
By 2011, Apple was the most valuable company in the US. Should it not have been allowed to start selling watches, home stereo equipment and AirPods?
Is this true? My recollection being a netflix DVD customer is that brick and mortar rental stores were dominant until netflix streaming upended the DVD rental dynamic. Would love to see some data on it though!
https://hbr.org/2011/04/how-i-did-it-blockbusters-former-ceo...
https://www.cnbc.com/2020/09/22/how-netflix-almost-lost-the-...
Would the massive subsidization of YouTube’s costs not count? There’s nothing physically preventing a competitor from buying storage and compute, but it’s pretty clear that it would be extremely difficult to compete with YouTube if every user has to pay the actual unit costs to upload, convert, and play videos.
That's on the user side, though. Like AWS and Azure, chances are GCP/Google's internal hardware division bills internal teams similarly to external customers (especially as Sundar continues to demand cost cuts). Of course users themselves can't go and make their own website and experience the same economies of scale that only big companies can achieve, but if competitors come in (like ByteDance) there's nothing stopping them from stealing marketshare from YouTube if they invest in some colocated servers in key markets.
Having worked on such large conglomerates it’s closer to friends & family discount than from large external company discount.
There’s also transfer pricing and other accounting ways to improve numbers.
This is so widespread that one such company had <company color> dollars (internal purchases) and green dollars (external customers) pricing.
Every major company operates the same way.
When Disney+ "buys" the right to stream the latest Marvel movie from Marvel Studios, the company as a whole has to back out the revenue so not to count it twice. But still show Disney+'s profit.
Disney+ buying from or selling to Marvel would not show up as profit or loss for Disney.
Making up numbers:
It costs Apple $800 to make a MacBook. It "sells" the MacBook to Apple Retail for $900 wholesale. Apple corporate now has revenue of $900 from the Mac it sold.
Apple Retail sells the same Mac to the end user for $1000. Apple Retail now has revenue of $1000.
Apple Corporate claims revenue of $900
Apple Retail claims revenue of $1000.
Apple claims a revenue of $1900. But it has to back out $900 of revenue so it doesn't double count it.
Apple wants to show that Apple Retail is profitable.
I used Apple instead of Disney in the example because it is much simpler to show one item being transferred from Apple to Apple retail than it is to show a group of movies being transferred to Disney and Disney selling a subscription.
Historically the concern was that railroads might get into say coal mining by charging competitors more to move coal than they charged themselves.
Microsoft for example could have spent unlimited money on IE and the justice department wouldn’t have cared as long as it was treated as a business investment. However, leveraging the ability to change windows was seen as problematic.
Long term subsidies can also run into issues, but it’s believed YouTube is inherently profitable.
But that still isn't logical. Are you really going to say that a company shouldn't be allowed to use resources to create a new product? Should it have been illegal for Netflix to use profits from shipping DVDs to later bundle video streaming?
Should it also be illegal for companies to "subsidize" open source contributors? Should it be illegal for YC and other VCs to "subsidize" money losing startups based on previous successes?
How long should a company be allowed to spend money on money losing products to gain a foothold? Microsoft probably lost money on the first 2 generations of the Xbox. Should that be illegal? Should Microsoft not be allowed to throw billions at ChatGPT?
And the critical thing you don’t seem to understand is resources aren’t the same thing as money. Suppose Intel decided to change their CPU design so only their compiler would have access to the full instruction set, that’s not something money can buy. Anyone can bring lots of money to the table, only the monopoly can leverage the resources associated with that monopoly.
As to spending money, again it’s not about gaining a foothold because anyone can spend money. The issue is if someone is subsidizing a business to maintain their monopoly rather than operate a profitable business. Google didn’t get into fiber because they wanted to operate a profitable business they got into fiber to defend their websites. They didn’t cross the line, but you can see how long term subsidies of different business could be seen as desirable/problematic.
And just in case you haven't noticed, Intel isn't exactly doing great against AMD these days and doesn't even have a monopoly on chips with the x86 instruction set.
Do you think its also illegal for the cable companies that originally lost money on cable infrastructure and to subsidize the cable infrastructure to sell subscription content? DirecTV probably wouldn't have made money selling satellite internet. But it was able to do it cheaply because of its content delivery. Every company leverages its resources to have margins of scale.
Even HN can only exist as a free service because it leverages the resources it has from the rest of its business. No one pays dang to moderate HN. The moderation is subsidized.
People get away with breaking the law constantly, you can do 57 mph in a 55 every day for 60 years and never get a ticket. Shipping Windows with a copy of IE was like doing 57, nominally illegal but not something they would have gotten in much trouble over. But like a speeder they pushed even further and got slapped down over it.
Intel has had an X86 monopoly at various points. Doing something when they aren’t a monopoly can be legal when doing the exact same thing as a monopoly is illegal. Whether they currently qualify as a monopoly is debatable, but you don’t need a 100% market share to be considered a monopoly. AMD having a 20% market share is right at the edge of where Intel would be considered a monopoly but Intel has a great deal of market power.
> Microsoft for example could have spent unlimited money on IE and the justice department wouldn’t have cared as long as it was treated as a business investment. However, leveraging the ability to change windows was seen as problematic.
There is no confusion. Nothing came out of the DOJ case with respect to MS bundling IE.
Intel has never had a monopoly when it came to x86. IBM insisted from day one that were two suppliers for x86 before it even agreed to use the first chip.
Making it more difficult to install competitors is “leveraging the ability to change windows.”
They very much lost in large part because of a video shown in court of how difficult to install a competitor. They even lied about it by editing their own video which skipped steps showing they knew that behavior was problematic.
As to Intel’s x86 monopoly. Intel paid 1.6 billion in response to a monopoly lawsuit, and the EU fined them $1.45 billion and ordered it to end its customer rebate program. So they where definitely treated like a monopoly.
They have even swapped peoples browsers with security updates before. Though more recent examples like KB 3135173 seem like a simple mistake, it’s more provocative when such things were common.
“Microsoft later submitted a second inaccurate videotape into evidence. The issue was how easy or difficult it was for America Online users to download and install Netscape Navigator onto a Windows PC. Microsoft's videotape showed the process as being quick and easy, resulting in the Netscape icon appearing on the user's desktop. The government produced its own videotape of the same process, revealing that Microsoft's videotape had conveniently removed a long and complex part of the procedure and that the Netscape icon was not placed on the desktop, requiring a user to search for it. Brad Chase, a Microsoft vice president, verified the government's tape and conceded that Microsoft's own tape was falsified.” https://en.wikipedia.org/wiki/United_States_v._Microsoft_Cor....
Also I had forgotten about this bit: “Later, Allchin re-ran the demonstration and provided a new videotape, but in so doing Microsoft dropped the claim that Windows is slowed down when IE is removed.”
https://corporatefinanceinstitute.com/resources/management/m...
FWIW, the reason you might be a bit confused--though I do not think this excuses the time you're spending here--is that Microsoft did not end up broken into two because they appealed that part of the ruling and "won"... but that court of appeals actually re-affirmed that Microsoft's actions were illegal!
https://www.computerworld.com/article/2582620/appeals-court-...
> The appeals court, in its decision, specifically rejected Jackson's order that Microsoft be split. "We vacate the judgment on remedies, because the trial judge engaged in impermissible ex parte contacts by holding secret interviews with members of the media and made numerous offensive comments about Microsoft officials in public statements outside of the courtroom, giving rise to an appearance of partiality," the decision said in part.
This is essentially akin to someone being convicted of something and then appealing the decision to an appeals court that decides "OK, the defendant clearly is guilty, but the chain of evidence used by the district attorney was compromised and so none of it is admissible; thereby, you cannot send them to jail, even though we know they are guilty".
(edit: Also, there were, in fact, serious ramifications on Microsoft for this loss... Bill Gates largely had to step down as CEO due to having gotten the courts so mad at Microsoft that it was almost broken apart, the IE team had a difficult time hiring anyone due to the shame and Internet Explorer ended up on life support <- this is often blamed in Microsoft "not caring as they were a monopoly" but that is far from the truth, and the resulting chaos for Microsoft really allowed Apple an entrance with Steve Jobs back at the helm to retake a large market segment. I did a talk about The Fall of the Roman Empire that covered some of these issues, but more detail on IE is likely a nonsequitur.) https://vimeo.com/310654342
Job first came back to prominence with the iMac, then really took off because of the iPod. Microsoft didn’t fail in the music player market for lack of trying, they had multiple failures - first Plays4Sure and then the Zune. It was execution failure. The same with the phone.
Microsoft was never able to compete with an integrated player like Apple except for the Mac.
Aka the use of private Windows API’s by IE was explicitly banned in the decent decree.
The initial case which MS lost would have resulted in splitting the company, on appeal and by the time of the new trial (which they also lost) the government changed its mind into no longer seeking to break up the company. But that was on the government not a legal victory by Microsoft.
> However, the DOJ did not require Microsoft to change any of its code nor did it prevent Microsoft from tying other software with Windows in the future
“H. Starting at the earlier of the release of Service Pack 1 for Windows XP or 12 months after the submission of this Final Judgment to the Court, Microsoft shall:
1. Allow end users (via a mechanism readily accessible from the desktop or Start menu such as an Add/Remove icon) and OEMs (via standard preinstallation kits) to enable or remove access to each Microsoft Middleware Product or Non-Microsoft Middleware Product by (a) displaying or removing icons, shortcuts, or menu entries on the desktop or Start menu, or anywhere else in a Windows Operating System Product where”
https://www.federalregister.gov/documents/2001/11/28/01-2949...
the only way to achieve the above is to change windows code.
The API issue wasn’t one of code it was one of documentation:
“The proposed settlement required Microsoft to share its application programming interfaces with third-party companies and appoint a panel of three people who would have full access to Microsoft's systems, records, and source code for five years in order to ensure compliance.[28]” https://en.wikipedia.org/wiki/United_States_v._Microsoft_Cor....
It publishing API’s may seem like a strange remedy but it was a common tactic they used to limit competition. Netscape for example couldn’t release Netscape for Windows 95 at release because while the IE tea, had access to API’s they weren’t providing them publicly until release.
In the sense that there was ultimately a comparative slap on the wrist remedy, in part because of a favorable-to-Microsoft change in administration between the findings of fact on antitrust violation (in which bundling IE was a factor, but far from the sole factor) and the imposition of a remedy for those violations, yes, nothing much came out of it.
Also, U.S. vs Microsoft wasn’t limited to Netscape vs IE several other companies were involved. The first Prohibited Conduct was in terms of OEMs:
“ A. Microsoft shall not retaliate against an OEM by altering Microsoft's commercial relations with that OEM, or by withholding newly introduced forms of non-monetary Consideration (including but not limited to new versions of existing forms of non-monetary Consideration) from that OEM, because it is known to Microsoft that the OEM is or is contemplating:” …
“B. Microsoft's provision of Windows Operating System Products to Covered OEMs shall be pursuant to uniform license agreements with uniform terms and conditions. Without limiting the foregoing, Microsoft shall charge each Covered OEM the applicable royalty for Windows Operating System Products as set forth on a schedule, to be established by Microsoft and published on a web site accessible to the Plaintiffs and all Covered OEMs, that provides for uniform royalties for Windows Operating System Products, except that:”
Etc
https://www.federalregister.gov/documents/2001/11/28/01-2949...
> Microsoft for example could have spent unlimited money on IE and the justice department wouldn’t have cared as long as it was treated as a business investment. However, leveraging the ability to change windows was seen as problematic.
And how is it related to anything Google is doing?
What in the hell is "market rate" for compute? Google bought a zillion CPUs and hard drives and put them in data centers that they pay for. There's no "market rate", they own the infrastructure. They can do whatever they want with the equipment they own. A division using another division's spare compute capacity isn't in any way an anti-trust violation. What you're saying makes no sense at all.
“Amazon EC2 Spot Instances let you take advantage of unused EC2 capacity in the AWS cloud. Spot Instances are available at up to a 90% discount compared to On-Demand prices.” https://aws.amazon.com/ec2/spot/
As a monopoly Google could sell the same capacity YouTube is using on Google cloud services and YouTube could then buy that capacity. If YouTube was treated as any other customer it’s fine, if YouTube got preferential treatment it’s not fine.
I doubt anyone is going to take them to court over this but it is likely illegal.
The problem comes from Alphabet having cheap compute left over from their search monopoly. In 1982 AT&T was broken up because it was using its local monopoly when competing in the long distance market. Again for 99% of companies it would have been fine, but monopoly = special rules.
And we shouldn't forget that, in the US anyway, it's not illegal to be a monopoly. The "antitrust" laws are about unfairly leveraging your monopoly position.
Where I would try to nail Google/Youtube for with anti-trust is the merging of Music and TV/Movies into Youtube as well as the reliance of Google ads on Youtube. Both of those are incredibly annoying to deal with as a competitor and give Google the ability to corner the market on multiple fronts. Roku has highlighted this themselves, Google leveraging Youtube to make brand deals that other competitors cannot.
Are you really saying that no company should be able to create a new product based on resources it already has?
Why stop at compute? Why not make it an "anti-trust" case that Apple leverages its same operating system and chip design across multiple devices? Or Amazon uses its same logistics network to deliver more than just books?
Your argument is valid, but scale matters. Think of a game of Monopoly; you can usually tell who's going to win well before the game concludes, and you can guess which players will lose well before that.
But let's not pretend that the little guy wants to go public. If they do, they are naive. Out of all of the companies, that YC has invested in, maybe 5 have actually gone public. Every startup knows the game is to get acquired.
The fact is that some things need scale to succeed. OpenAI for instance was never going to be able to afford the compute it needed without being subsidized by MS.
Both companies had a first mover advantage by deploying a more scaleable technical innovation in a somewhat stealthy way. OpenAI is subsidized as you say, but StabilityAI promptly ate their lunch int he image production field and is hoping to do the same with LLMs.
A "successful" business makes more than it spends so it can be an ongoing concern. Any company can sell a lot of dollars for 95 cents. Youtube was never a "successful" business with a meaningful revenue stream before being acquired by Google.
> Or how did Google come to exist, when the market was dominated by (then-) huge firms like Yahoo, Lycos, and Altavista?
You realize you're arguing against the need for government regulation. This is the market working as it should. But Google became a success initially because Yahoo used it as their "search engine". Google initially offered to be bought up by Yahoo and Yahoo balked at the price. Google wasn't trying to go it alone.
You realize you're arguing against the need for government regulation.
I'm looking at some gaps in your logic when you assert that anyone who ever wants to go public is naive; when Excite (not Yahoo) declined Google's $1m buyout ask, they were already running as a public service under their own brand and had a plan for text ad sales in place. Certainly it was worth trying to develop a technology product followed by an acquisition before the more challenging (but perhaps less interesting) business of raising capital directly, but there was a widespread view at the time (among both tech and VC people) that their product was good enough for them to go all the way.
I'm not trying to debate the need for government regulation and don't know why you think this is some sort of 'gotcha'.
And who was going to fund them?
Which is why it's anti-competitive.
While it could be made open, I don't see any regulatory reason to. We want companies to reduce waste, and it doesn't make sense to require all companies to build public facing cloud, and integrate their flagships into it, just to open it up to their competitors. It would be like telling Apple that its factories reusing scrap aluminum to make iPhone cases is an unfair competitive practice.
I worked in WANOPS at Microsoft in the 90s, so my understanding of billing is literally decades old.
Google could rent this compute capacity to the masses, allowing for healthier competition.
They could, but no major corporation wants healthier competition. They want no competition whatsoever. So there is no reason at all for Google to do this, and every reason (from their point of view) not to.
So I'm skeptical that anything is actually "free" and, at a minimum, I'm sure has internal transfer pricing that has YouTube paying a reasonable value for the resources it uses. And I would furthermore assume that Alphabet management is satisfied with what YouTube's income statement looks like given those transfer payments.
It is hard to decide if this is unfair or not. It is true that Google doesn't offer this exact ability to external customers but most customers wouldn't want this type of very unreliable infrastructure. If there was a market Google or Amazon would likely offer it (basically a super-spot which has less guarantees)
E.g., yes they'll "pay" for compute, but they'll do so at cost. And they're also allowed to deploy a monitoring widgets to other orgs' data centers to find when they're at low utilization, and then argue "at cost" is actually near 0, because the compute resources were just sitting around anyway. And then have the search team agree because YouTube promises to build the new data API search has been asking for forever.
If you ultimately report earnings on the same P&L, USD is just monopoly money you trade around to get things done.
Team A is running a database and needs to buy disks with a certain amount of IO capacity.
Team B is running YouTube and needs to store tons of data.
You add up the IO + storage requirements from team A and the IO + storage requirements of team B and you buy disks capable of providing that combined amount of IO + storage. The resulting cost is much lower than the cost of buying disks for team A and disks for team B separately.
Is YouTube viable as its own company at the scale it operates? I have to imagine the infrastructure costs for YouTube would be extremely high and would likely cause YouTube to downgrade quality or try even harder to put a subscription behind even 1080.
Now obviously contacts can be signed and I highly doubt they would be paying normal prices in any cloud environment given this hypothetical situation. But it wouldn't be the same as it would be now.
Does YT need Google's backing in order to survive because they don't have the resources strictly in-house to build and pay for infrastructure?
Disclaimer: Not an Economist, not even close.
If in the unlikely case they spun it off and another competitor emerged, would YT's scale make it hard to compete with this new entrant? Read: Red tape, bureaucracy, etc, etc.
Essentially back to where they were as an independent company when they didn't have Google to "protect" them...
The inverse case is what usually comes to mind first though:
Would the company wither and die if not for an acquisition? Think about Sprint+TMobile. Until the deal, Sprint was in bad shape, second to Verizon with outdated tech - in the sense that emerging technologies were based more off of an evolution of GSM than CDMA.
If TMo doesn't acquire them, Sprint probably goes "poof" and their customers flee to the remaining carriers, further entrenching their positions.
This seems like it is usually the card European regulators tend to play, worrying more about market dynamics and how that will affect costs rather than just what people pay for service.
There is only so many hours in the day and so many eyeballs watching, that is your upper bound. Bandwidth like that is not available to mere mortals, you would have to colocate with a lot of ISPs all over the place. Would take years to setup, not really even a question of money.
In the case of Netflix their content library fits inside one such colocated server, which can (or pretty soon will) do 1tb/s. Hundreds of thousands of streams per box are within reach, especially at YouTube potato resolutions.
In the case of Youtube while the long tail is very very long, I bet the most popular content of the week would fit into a similar sort of server.
So the serving portion at least is an endeavor on the order of magnitude of say 1 billion dollars a year.
No idea how to estimate letting anybody upload unlimited amounts of video of unlimited length and transcoding it into a dozen sizes. I think they did that early on to solidify their moat and it seems like a huge waste.