I think the scale where it makes sense is at companies where paying $780/month to an existing provider is easier than all the paperwork and internal regulations around allowing people to provision on prem equipment or use an untrusted provider.

Consider the item from yesterday about AWS Outposts: https://ahl.medium.com/aws-outposts-2021-stories-for-folks-w... .. the markup is absolutely mind boggling, but there are companies who pay it.

AWS is the master of having a pricing model that feels nice to most indies and small/medium companies but that scales up extremely effectively to capture huge amounts of enterprise spend at the same time.

I mean, you're not wrong.

For example, I know many furry artists who open their commissions and within an hour get 20 solid offers to pay them. They can't possibly do all the work, so they just lose business.

That's, uh, not how capitalism works! Companies like Amazon know, all you have to do is charge 10x more to capture all of that value. Sure, it sucks for 18 other potential customers, but these furry artists have tens of thousands of fans who will enjoy the work regardless.

Here's a really interesting pricing model your friend might be able to use to capture more value (and not make the community mad).

https://kevinlynagh.com/notes/pricing-niche-products/

Great example of the power of auctions! It's amazing to see the Vickrey Auction come off the blackboard and enable an indie creator to connect with their community sustainably. Markets really are all about moving information so we can coordinate activity.

Wouldn't it be awesome for Kickstarter / Patreon to add this as a feature?

Reading more of it, it looks an awful lot like the business model behind YCH auctions-- Your Character Here. That's a very successful model, and it often results in prices $400+, which is about 2-3x more than what artists often charge otherwise.

That said, YCH is a more traditional auction over a single piece, not a Vickrey auction over a specific supply. According to this model, a Vickrey model would be:

- Scope the work to a specific standard that the artist specifies, in terms of colors, quality, characters, background, etc. That way, the bid is on a roughly equivalent amount of labor per piece.

- People put in (blind) what they're willing to pay

- When demand is exceeded, the nth+1 highest is what everyone pays.

Very fair!

I'd be curious if there'd be a way to price something more custom; for example, putting a price on the number of hours required for each feature (more color or shading, more characters, etc.), how would that affect such a pricing dynamic?

Does such an auction hold under the modification made for multiple participant?

Generally if you intentionally overbid you end up paying the otherwise highest amount discouraging that kind of action (ie you may as well put your max amount).

But if the top 10 pay the price of the 11th bid there’s a perverse incentive for me individually to just say I’m bidding $1 million because I know that most other players are likely to not be smart enough (at least in the short term) to bid strategically and thus I’m guaranteed to get the item at a steeper discount without having to reveal (or bother thinking about) my true max price for the item.

If you do that, there's a real risk you'll end up having to pay more than your max price. Say your max price is $2500, but you bid $1 million. Then it turns out 10 others have a max price of, and bid, $10000. You are now on the hook for $10000, significantly more than your actual max price.

That's a really good idea! I'm going to send this to the people who make https://commiss.io, thanks!

You're assuming that 2 of those 20 potential customers would pay 10x, when it could be that 20 of the 20 would balk at 1.1x.

Of course that requires some market analysis not a straight up "let's charge 20x". But when talking about limited capacity in general it makes sense that you overcharge to capture as much value as possible. Some customers are always willing to pay more so if you somehow know or can guess the distribution of offers it makes sense to capture everything from the most valuable buyer group down to whatever your capacity can cover rather than a random collection in the middle.

In the case of the M1 it might make sense to charge at these levels especially for customers that aren't interested in using anywhere near the full value of the product. So if a company or developer estimates it will only use the equivalent of 50% of the Mac's price it would be a better deal than purchasing the thing outright and also pay the associated operational costs. As always you have to know your market if you want to maximize the profit.

Not a furry artist but similar vein. There are genuinely a surprising number of people who are willing to pay 5 to 10x the market rate for commissions.

That’s first step of capitalism.

When someone makes crazy profit, you’ll have fast followers.

More artist will switch. Prices will drop. At some point some artist will drop out and a rough equilibrium will occur.

Then one of the artist will lobby a law that limits the number of furry artist.

Prices go back up due to restriction.

People complain at high cost of furry art. Licensed artist will lobby a new tax on all artist to subsidize the cost of furry art.

Now non furry artist are paying the furry artist.

Customers get confused on using tax vouchers on Furry artist.

Non furry artist who complain are now regularly attacked for not supporting the Arts.

Ouch. That's cynical. But, sadly not too far from the truth.

But, what's the alternative? Capitalism without human nature?

Everything in story was fine until government got involved.

Capitalism increases efficiency by being ruthless. Better, cheaper, more or die.

Government and regulatory capture screw it all up.

But what about the trees, treeman79?

The problem is capitalism optimizes for profit, not life.

Some regulation.

Not allowing Dumping toxic Waste in river is one thing.

Banning ride sharing In cities to protect taxi companies is another.

>AWS is the master of having a pricing model that feels nice to most indies and small/medium companies but that scales up extremely effectively to capture huge amounts of enterprise spend at the same time.

I wonder if there are any other industry that does the same. Most ( if not all ) of the time pricing dont scale to both ends. But AWS pricing seems to be the outliner here.

Taschen is a publisher that has somehow pulled this off in what you'd think would be a highly commoditized space: books.

They sell normal sized versions of most of their books for $20-$80, some slightly premium ones for up to a few hundred dollars, but then have these large coffee table sized books in limited editions selling for $10k+: https://www.taschen.com/pages/en/catalogue/photography/all/0...

Normally when I shop around for other cloud providers, I see AWS’s “pay us more but everything’s under one roof” strategy in action: Other cloud/infra providers may beat them on price, but IME it’s often like 10-20% at most. Maybe 2x in some cases.

It could be that there are companies for whom a contract with MacStadium is too much to handle, but for anyone who isn’t a supergiant enterprise, I just can’t see how an 8x markup over the competition makes any sense.

I’m sure at this point some enterprise customers are just hard-wired to reach for the AWS solution on everything. But I also wonder if this way of thinking has an actual hard limit somewhere.

I'm not particularly experienced at the enterprise level but I sometimes roleplay the conversations I expect they have in my head ;-)

I can just picture a department needing to do some CI on a Mac, having access to a certain (high) budget on AWS, and going with Amazon's offering rather than put together the business case to authorize the MacStadium spend. Or if it's under their approval amount, figuring out who's responsible for the thing when inevitably staff change around, etc. If the company doesn't bat an eyelid at another $800/mo on the AWS bill, I can see why many people would be tempted to click the button and get on with their work day :-)

I would appreciate people who live in this enterprisey world to fact check me on this as my experiences in this matter are rather limited..

I can confirm that if you have a large contract already setup it is much easier to add to it than go through the approval process for outside spend.

I can click a button that costs us $5k and nobody blinks an eye, but I have to request permission to spend $100 to do the same elsewhere.

If you have a giant pile of data already in AWS, that you need to use on a Mac for a short time, it may make more sense to rent a Mac next to all your existing data, rather than paying exorbitant transfer fees to get the data to a Mac elsewhere.

AWS's transfer fees are a massive lock-in.

That's a big assumption on how EVERY company that needs an M1 and uses AWS uses it for

The most common use case for a Mac in a business setting is for building iOS applications.

You may have to define business.

Account managers, web developers, python developers, Julia developers, sysadmin, netsec.

All those areas are riddled with Mac.

But none of that code runs on Mac alone. Even if they're developing on a Max their python or Julia or whatever code that they want to run on the cloud will run on Linux. The only real use case out there is building/testing iOS/Mac applications. The only other one I can think of is a Mac only video editing or animation software that you'd prefer to render somewhere besides your laptop.

Although to me-and-you and to an extent SME's it may seem crazy to buy one over running it on-prem or at other companies to large organisations it makes perfect sense and the cost is just a drop in the ocean compared to their normal AWS monthly bill. Lets not forget AWS will only invest time if there are customers who will actually use it.

A few points from the top of my head:

* Large enterprises won't pay the costs listed publicly, they will each have their own contract written up with AWS.

* Contracts - it can take YEARS for new companies to be added to the approved vendors list.

* Frameworks will already be in place for how resources are managed and accessed securely.

* Enterprise support agreements will already be in place.

* Physical security that is regularly audited

Given the ridiculously low power requirement of m1 chips if anyone wants to use these probably in small scale capacities they should just buy it. The power and maintenance won't be that much if a hassle (10 of these would use less than 300w) and you would get warranty too. I think the mac mini m1 was designed (lack of) by apple to replace current mac mini racks but ideally this thing can be as tiny as a 3rd gen apple tv and a rack of 10-20 could be as small as a mac pro ...

Well thats what we need from Apple next. The Mac Rack of 10 M1 chips (aka. Mack). Not sure how useful that would be for me personally though. My reasoning being, if 1 chip takes 30W, why not double the number of chips or do something to the chip to get more from it.

I wouldn't be surprised if Mac Pro version of the M1 has everything turned up to 11.

it's going to be interesting with the 16 inch macbook pro the imac and mac pro. of course apple has the full licence with ARM to do almost anything they want but based on my understanding the X ARM cores the 7 series and 5 series can be clustered. My guess would be apple moves to a cluster based design eventually like in m1 but with more cores for higher end machines but right now i'd say just a chiplet version of current m1 chips would still be a great option with higher frequency. A 4 or 8 chiplet design would still be only 150w or so with gpus disabled or somehow using all gpus as one. This is the strategy amd has gone with obviously with chiplet without gpus but I think this also makes sense for apple to reuse its designs.

don't hold your breath. There are good reasons the X-Serve line doesn't exist any more.

Do those reasons still exist in 2021? What are/were some of them? (Genuinely curious, not playing devil's advocate)

I have no idea why it was canned, but it was a very expensive server with a very expensive SAN solution. I think it just didn't compete well with other enterprise class machines on price.

People are willing to pay more for a Mac Laptop/ desktop because the overall experience is nice. The screen, the construction, keyboard (save the shitty one), trackpad, Retina display, high speed SSD, etc all significantly affect user experience.

When the primary use case is sitting in a closet and serving up bits quickly, it quickly boils down to bang for the buck where the Mac has always lagged. The M1 could very well change this though. Particularly since power use is significantly better.

Also, when the Xserve was around, there weren't nearly as many iPhone developers looking for CI solutions. Seems like now there would be a pretty healthy demand for that alone.

If you have, for example, a profitable cross-platform mobile app, but your developers primarily use windows/linux and test on Android, then it might make sense. If all you need is to spin up a mac for the occasional build/test/deploy, then paying $85 per month is easier than having to maintain an M1 in the office that no one is going to regularly use, especially if your team is distributed and/or remote.

In many large companies, operational costs are often budgeted differently from capital expenditures.

The workflow from approving a $85/month cost can be significantly easier than approving a $1,099 purchase.

Also, renting an machine from a 3rd party might allow you to avoid dealing with a dysfunctional IT department.

AWS’s margin is Scaleway’s opportunity, you might say.

Normally yes, it should be, but I don't think it's going to happen. People have been (1) miscalculating the costs of AWS, (2) realizing they are relatively high but deciding to stick to them for various reasons for a long time now. And because Amazon managed to capture many companies with their marketing, in many cases it's practically impossible to get out.

Yeah. It was just a snarky play on the famous Bezos quote, "Your margin is my opportunity."

This comment is on most AWS stories. Bandwidth charges alone are usually significantly higher than you can get elsewhere.

My own guess - a bit of a niche product not super easy to scale do they set a rediculous price. If the governance wins are there - they may still sell some - they rarely do stuff w out customer demand. At least in govt contracting - if you are willing to jump through all the hoops, some insane markups exist.

I would assume that Amazon's offering is targeted at customers who already using other AWS services. So even if a Mac Mini costs $500 per month more, it's really not that big of a deal considering how much the bureaucracy of signing up with a new hosting provider would cost in a big company (e.g., legal, infosec, ...).

Ohh yes. We had built a small tool that business people had to use, in form of a questionnaire, that went through about 20 topics. Depending on the question/answer it was an immediate "nope, must-have, don't sign a contract".

Ironically this (nope-case) was always the case if the provider ran on AWS.

Few things on AWS make sense financially. They're just leveraging the power of lock-in™. Why save money on AWS when you can save money on employee salaries?

But that would be an asset, not an expense. It depreciates over time and your accounting department must track it somehow in the books.

Ok, maybe you could sign a leasing for the mini; and still be cheaper than AMZ but... your company would need some credit scoring and your finance department should talk with someone...

But for $26/day it gets buried for free in a bill that needs a cloud architect to decipher.

Makes sense if you only need it for an hour every day to do a build and/or run a QA automation test. The billing on it is per-second.

Worth noting that they require 1 day minimum to play ball, so kicking the tires is not a per-hour option. There is also no spot market for them, and no reserved instance pricing either.

It seems like AWS has really replicated the Apple experience in more ways than hardware.

I understand that the 1 day requirement is from Apple and thus AWS requires that minimum too.

Why did Apple require a 1 day minimum?

>https://www.youtube.com/watch?v=jZJnrKjfA3s&feature=emb_titl...

Start watching the video, thinking no baguette?? 1:03 here it is!! Thanks France, love your sense of humor!!

Those are MRPs. Amazon will be giving heavy discount to large customers anyway