Hacker News Reader

Comment by mattacular | Hacker News Reader

Don't forget that Amazon dominates a lot more than just retail on the internet. With AWS they effectively run large portions of it as well.

View on HN

They run large portions of it as essentially a gargantuan hosting company, in data centers in a few fixed locations.

They buy a lot of dark fiber, and IRUs, and wavelength services from carrier-of-carrier type big ISPs, but by no means do they control the internet.

But they are not in the same business of long distance OSI layer 1 ISP operations as CenturyLink/level3, zayo or Verizon are. Where I would be really worried is if I saw amazon making a serious bid to acquire a company similar to Zayo, NTT, KT, France telecom/opentransit, or Telia.

It's a good time for Amazon or Google to buy a Telco, Telcos valuations have lost between 25 to 45% since the beginning of 2020 [0]

I guess that a state owned company would willingly be sold by its owner because states urgently need money.

For example Orange sa (ex France telecom) is owned at 25% by the French state and French state just had passed a budget which is financed at half by debt (2021). Selling a company for $b30 would provide some relief in 2021 budget.

[0] https://www.infrontanalytics.com/fe-en/FR0000133308/Orange-S...

There's no point in buying a Teleco today, with 5G any of the FAANGs can create one with the money they have. Infact Google, Amazon, Microsoft and Facebook have proven experience in web scale, so 5G will be right up their alley

That's true in principle, they could buy directly at Telecom manufacturers, rent antenna space and use subcontractors to install base stations, but there are counter examples showing they are not so good at this business.

For example Google has tried to provide free Wi-Fi, at the time some people were afraid it would be some rehearsal for a Wimax deployment, it never happened.

8 years ago, Google made a lot of noise with their high altitude platforms (their so called balloons). They met some Telcos, but it went nowhere.

Long time ago Google had also a project to provide phone calls with interesting fees. It was a MVNO in practice but it never got traction even if Google had access to an incredible number of Android users.

Phones call with interesting fees? Do you mean Google Fi which I know many users of?

If you think "5G" will replace medium and long haul DWDM fiber networks, as a backbone, you have some very great misperceptions about what real internet infrastructure looks like.

I never said that, I don't think you really understand how 5G works. The big 3 cloud already have extensive experience in creating DWDM fiber network as backbone, it's not really all that different from internet stack anymore.

But none of them have their own ROW to do it, except around small specific regional campuses like Hillsboro. They have dark fiber leases, IRUs, and wavelength circuits. Show me a big FAANG owned and controlled cable from western OR or WA to Colorado, for instance. Doesn't exist. Show me a FAANG owned and controlled path from Hillsboro to Sacramento and the bay area. Doesn't exist. They have IRUs in infrastructure built by someone else (Zayo, level3).

Show me one FAANG with anything even 5% of the outside plant wholly contolled fiber route miles of the size and scope of the railway ROW sprint, qwest and level3 networks. Or legacy wiltel.

The only real exception, Google has successfully made a big deal of itself in submarine fiber by throwing its money and weight around.

Of course they don't do it at the scale old telecos are doing it, doesn't mean they are completely new to this kind of thing. Doesn't really take a genius to do this anymore. AWS directly invests in third party companies to lay out intercontinental dark fiber for them, wouldn't take them much to do it themselves if they want to do it on the scale of a teleco, it just isn't a good financial decision to do so at the moment, not a lack of technical expertise.

You think Zayo is a "old teleco"?

You are just deflecting the main point of the argument now. Companies like Jio have already proven that all you need is to throw money at this problem and it can be solved. You just need to make sure that the money you invested can be earned back.

With all due respect I sincerely doubt you have ever had the equivalent of "enable" or "configure" access on the core routers of any significant ISP. Not anything in the top 1500 worldwide by CAIDA ASrank size and customer AS, peer AS measurement scale.

Using jio as an example of "why don't the faangs just build their own massive ISPs at OSI layer 1 by throwing money at it" is so dreadfully wrong and misinformed I won't even dignify it with a further response.

Isn't that kinda like saying that the electric companies do the same thing, though?

They kind of do, which is why alot of big tech should be highly regulated like utilities.

I feel like saying that AWS should be regulated like a utility because lots of businesses use their products is a little sill. They have heavy competition in basically all of their markets, their moat is their breadth of services and they’re constantly at risk of being unbundled by a competitor that has more time to devote to a single product segment. They have persistent natural competition with colo-ing that keeps their prices from straying too far from commodity hardware and an ops person.

Like if you’re looking to regulate a market this seems like one of the last you’d peg as not being healthy.

There are tons of niche hosts that have sprung up recently, a lot for developing and serving AI/ML applications.

Amazon was there first, but it's getting cheaper by the day to buy a bunch of hardware, stick it in an increasingly cheaply electrified building, run a bunch of fiber to it and slap some virtualization software on it.

AWS wouldn't be my first choice to regulate, but at this point they can turn any competing business into a feature of AWS without breaking a sweat. I know for sure this stifles innovation because I've watched people abandon promising ideas because of that fact.

I’m under the impression there is more regulation because they are a utility company.

AWS has varying degrees of vendor lock-in, which electricity companies generally don't have.

Electric companies can switch off your electricity, and not much else.

Amazon can read your secrets.

I know that in practice nobody at AWS reads customer secrets and if someone does, it will be logged somewhere but I always find it funny that you can encrypt things in AWS with secrets fully managed by AWS. It's just a checkbox to check the box for encryption in technical requirements but it doesn't really matter if you really care about privacy when the keys are stored in a very safe box accessible by AWS and the USA state (patriot act).

> I know that in practice nobody at AWS reads customer secrets and if someone does, it will be logged somewhere

Can you share how you know this to be true?

I certainly hope it's true, but I have no way of knowing myself.

I actually don't know, you are right. I have some trust in AWS but I would never put a real secret in AWS in clear text.

True, that’s why in some markets the electric companies have been split into transmission companies and delivery companies. You buy power from the delivery company and the transmission companies are paid to maintain power lines.

That's effectively how it already is though: there's one market for connectivity, and a separate market for cloud computing. If you don't want to build on Amazon's cloud, you can build on Azure, GCP, or one of the smaller competitors.

Isn't that a scam from Enron that we are still living with?

In New York we have almost as many ESCOs as there are shady looking prepaid phone cards. Maybe they save you 0.001% over the real electric company if you cash the check for 78 cents they send you every six months.

I thought the point of that was financial engineering. Used to be if you wanted to shut down a coal burning power plant the power company would be sued by creditors. Now the creditors can sue a shell company with no assets and get nothing.

It is good that coal burning power plants are being shut down, but rather than having an engineering or environmental based plan they do it this way. There will be plenty of money for solar roofs if they government is paying but private money will not be there for other investments if it doesn't have to be paid back.

Yes but that's why they're classified as a utility

That’s different, though, in that they have significant competition and the users don’t care at all. They have a lot more ability to affect prices on retail sales.

Don't underestimate how Google is becomming the storefront of all retail on internet.

Google can now dictate which webshop is shown first in search results on a product level.

I know retailers are sick of this. They need Google to be found on the web but they don't like how Google is destroying retail.

There isn't really a strong narrative that Google Shopping is any kind of threat to Amazon or Walmart. Nobody knows if it's going to catch momentum, or if Google will just kill in a few years.

There's a huge opportunity for Shopify to launch a slick shopping service that puts a (much nicer looking) Amazon-style interface in front of all the various shops. Combine that with Etsy's marketplace and you'd really have a contender.

They already do that.

Not well. I've never seen it and I spend way too much with Amazon (meaning I'm an ideal customer).

Independent aggregators are staging a minor comeback, thanks in part to the dismal failure of Google Shopping. I only ever buy PC components through PCPartPicker, I only buy games through IsThereAnyDeal. Third parties can offer real value in this space.

Reply on news.ycombinator.com