Jump to content

Sign in to follow this  
Krakin

A look into what Azure really costs the Xbox division

Recommended Posts

Hey, I could be wrong about a required subscription (highly unlikely) but i want to dispel this myth that streaming for FS2020 is going to cost MS too much money and force them to charge a monthly sub. Here's what Uncle Phil has to say.

 

Now keep in mind that he's talking about ALL of the games that are going to make use of the system. In the grand scheme of things, MSFS ain't that special. You can stop watching around 3:45 or continue if you haven't seen this interview yet.

  • Like 1

Share this post


Link to post
Share on other sites

The business analysis we see on AVSIM can be even more entertaining than the armchair software development opinions! I'll play too.

a) As far as I'm concerned the subscription thing is a pure AVSIM conspiracy theory. I think it started with people misunderstanding MS game pass.

b) MSFS2020 is barely a blip in MS's massive business. It's not like MS has created this infrastructure to generate profit from MSFS2020. I see MSFS2020 as a proof of concept, experiment on how there can be vertical/horizontal integration between Bing, Azure, gaming, etc. The industrial/commercial applications of a virtual MS world would dwarf the profits from a flight sim. Environmental, logistical simulations, etc.

MSFS2020 is not driving Azure; it's the other way around.

  • Like 4
  • Upvote 8

Share this post


Link to post
Share on other sites

It’s funny how everywhere azure is already - I only started to notice after the new game was announced, but pretty much the whole of the UK’s public sector uses azure in one way or another (the variant I came across was allocating and activating my work iPhone, but it also handles all the automatic software allocations on various different types of devices where I work too). It’s a million times quicker than when it used to be humans doing it (about a week’s wait, down to about 15 minutes). 

Share this post


Link to post
Share on other sites

I don't get it, maybe I missed the point but all I heard is "...there is a huge cost to Azure..." - that is the most vague and useless statement and doesn't answer your question.  You can say there is a huge cost to setting up any gaming data centre.  Microsoft Azure is a cloud platform used for tons of things, it's not exclusively for gaming.  It's designed to compete with the likes of Amazon AWS and Google Cloud Platform. Ofcourse the cost of setting it up is huge but that has zero relevance on what the cost for a single user (i.e. MSFS) of that platform is going to use.

I really don't understand why people are all over the internet saying things along the lines of "there is definitely going to be a subscription model".  Firstly, there are many ways to offset the cost of data centres, narrow minded people seem to jump to the conclusion that subscriptions is the only way.  Secondly, Microsoft themselves said they did not want to use a subscription model, which is smart and the correct thinking if they want this product to survive.  Thirdly, when people go around saying "subscription is the only way", then you are just asking for Microsoft to do that.

I know personally, this game will be of ZERO interest to me if it is subscription based, sure I may buy one month and then I'll be happy to see it burn to the ground because of such a stupid funding model, X-Plane 11 is good enough for me.  We've funded our flight simulators entirely on add-ons for the past decades, yet no one seems to mention that as a viable way to fund servers.

Everyone thinks they can solve Microsoft's problems when really it is none of your business to do so.  The only thing that is your business, as an end user, is whether YOU want a subscription model or not.  That's the only thing Microsoft is interested in, they want to know what people want, not what people think Microsoft should do.  You are not running their business, you are an end user of their service.

Edited by KillerKlient
  • Like 1
  • Upvote 2

1_logo2small.png

PVA1103 Robert Stevens

Share this post


Link to post
Share on other sites

Recently I've noticed TV ads in the UK advertising Microsoft AI.   I'm assuming this is Azure.

Share this post


Link to post
Share on other sites

There seems to be some strange voodoo being held with the word "subscription", as if saying it enough times will conjure a subscription model game. The reality here is that even though they have said they themselves would prefer it not be, the fact that they are unable to say that for sure it is not means that there is still negotiation going on within MS about how this project is positioned internally with respect to revenue.

The flight simulator market is not enormous. It is big, and clearly MS wants to dip into casual players as well, but FSX, released in '06, is reported to have moved only 1M units by '08. Contrast that with a game like GTA V, which moved 110M units. Let's say the game does really well and quadruples that by doing 2M units in the first year as opposed to 1M in 2 years. Now, Asobo is a relatively large studio, supposedly 140+ folks. And reportedly the game has been in the works in some capacity for 5 years. That's not the only project they have on their plate, but I imagine you could easily see 50 people assigned to the project. Development costs could very much be around 10M/yr, quite easily (a $50M budget for this kind of game is not out of the ballpark at all anymore). Suddenly doing just 2M units doesn't look that great, from a profit perspective.

And we haven't even dipped into the ongoing costs of a CDN backing terabytes of scenery data, the machine learning servers that will need to be scanning scenery data and generating models, tree masks, roof colors, building types, etc. Those servers need to keep generating that data with each new Bing Maps data update, so somewhat constantly. And in order for us to get all of this data, those servers have to live effectively forever. Bandwidth is far from free in the cloud.

Furthermore, we also know that Asobo and MS both plan to make MSFS a platform that they will iterate on. Which means ongoing development costs as well.

What can one read from the tea leaves here? That MS is likely trying to decide if this project is more of a technical investment or more of a traditional game project. If they decide it's more of a traditional game project, I think there's no way to avoid a subscription-like model, whether that means that future periodic updates are paid or it's a true monthly subscription, because otherwise the number of units a game like this moves just doesn't turn a profit at that level of expenditure.

However, if MS decides that this is more of a technical investment and position the project as a demonstration of what is possible for developers to do with Bing data and machine learning in Azure, then they may be willing to run the MSFS project as a loss leader, hoping to snag other large scale development projects (game or otherwise) that would need multi-millions per year in the cloud to achieve their goals. If anyone is positioned to do this, it is MS, but I don't think the community does itself any favors when it starts to project an entitled attitude.

-Matt N

  • Like 2
  • Upvote 1

Matt Nischan
Architect | Lead Developer

ZF67cij.png

Share this post


Link to post
Share on other sites

I know from my point of view, I'd rather pay $20-$50 a month for cloud gaming than pay $3000-$5000 on a depreciating asset every 5-7 years.

A mid range desktop with fiber could essentially get top notch performance results, without the crazy ever-adaption hardware costs.

Edited by ranold
  • Upvote 1

AMD Ryzen 7 3700X, MSI VENTUS 3x OC RTX 3070 8 GB, 32 GB DDR4 3600MHz, 1 TB NVMe SSD, 1 TB HDD, Sharp 43" 4KTV

Share this post


Link to post
Share on other sites
20 minutes ago, MattNischan said:

The reality here is that even though they have said they themselves would prefer it not be, the fact that they are unable to say that for sure it is not means that there is still negotiation going on within MS about how this project is positioned internally with respect to revenue.

You're probably reading too much into that panel response. In fact, glass-half-full personalities could look at that statement as a sign it's not going to be subscription based. It's not that guy's job to make pricing statements a year ahead of a game release. You had guys like Froogle coming from purely negative spaces (he had to take down his initial video response to the MSFS2020 E3 trailer because it was so irrationally vitriolic) and just looking for reasons to complain and find problems.

  • Like 1

Share this post


Link to post
Share on other sites

Thanks @MattNischan for adding some needed realism to these discussions. Have dealt with Microsoft professionally for over 2 decades, I can tell you they will support a high profile loss leader as long as their stock price is good. When their shareholders demand more profits, they will drop a loser product as quickly as they dropped FSX. 

PS...never forget it is never the engineers at Microsoft that decide which products are maintained, supported, or axed. It is the executives who are watching the stock price and profitability.

Edited by yurei

Share this post


Link to post
Share on other sites

I think this is proof that no matter what evidence you put forward there will always be people willing to speculate out their posteriors. Um...did we watch the same video? @MattNischan You are not arriving at any conclusions using facts. I mean using FSX as an example when you don't even know if your sales figures are accurate is just not wise. "And we haven't even dipped into the ongoing costs of a CDN backing terabytes of scenery data, the machine learning servers that will need to be scanning scenery data and generating models, tree masks, roof colors, building types, etc." Why even bring that up in light of what Phil said? Is the entire Azure infrastructure made for FS2020?

Also when companies see that their games aren't pulling enough profits due to under performance, they do one of two things. Eat their loss or go free to play if possible. Slapping a subscription on is one of the DUMBEST things you could do. Only of the hardest hard core among us would be willing to pay a sub for the sim. If you really think they are looking to dip into casual market with that strategy then they're sending the game out to die.

@carbonbasedlifeform is correct. You're reading too much into what was said. No dev is going to talk about pricing a year from release when the higher ups haven't released a statement. I do think they know enough about the cost of the project to allow them to say "we don't want it".
 

 

  • Like 1
  • Upvote 1

Share this post


Link to post
Share on other sites
2 minutes ago, Krakin said:

@MattNischan You are not arriving at any conclusions using facts.

I mean, neither is anyone else. Truly the only fact is when they release the price.

I'm merely speculating given the public sales numbers of FSX, and my knowledge of development and cloud computing costs, having been involved in large scale projects in both areas. I think people may not really quite understand the expense there. The kind of setup required for the application being developed by Asobo is on the order of hundreds of thousands of dollars per month in cloud costs, easily. I'm not even saying a subscription is a guarantee. My personal sense is that the faction in MS pushing the game is themselves trying to sell it to management as an investment. Am I reading into things? Yeah, probably, but I've been on that side of the table making the same pitch before.

All I'm saying is that if they don't go the sub route, it will be because of the way the application is positioned in the organization. I find the attitude that a subscription is downright unreasonable and/or unfair to be a strange attitude. I think, given the development, ongoing costs, constant streaming data, weather servers, continuous feature additions, and the like, that having the attitude that a subscription is a slap in the face is the same as wanting something for nothing.

There's simply no way that everyone buying the game once funds a huge cloud architecture and dev team for the next 10 years (the timeframe they said they wanted to support the game).


Matt Nischan
Architect | Lead Developer

ZF67cij.png

Share this post


Link to post
Share on other sites
25 minutes ago, MattNischan said:

There's simply no way that everyone buying the game once funds a huge cloud architecture and dev team for the next 10 years (the timeframe they said they wanted to support the game).

You do...not...know...what...you're talking about. You do not know how much Azure utilization is costing MS so stop talking like you do. You are completely ignoring what Phil Spenser has said.

  • Like 1
  • Upvote 1

Share this post


Link to post
Share on other sites
3 hours ago, MattNischan said:

There seems to be some strange voodoo being held with the word "subscription", as if saying it enough times will conjure a subscription model game. The reality here is that even though they have said they themselves would prefer it not be, the fact that they are unable to say that for sure it is not means that there is still negotiation going on within MS about how this project is positioned internally with respect to revenue.

It's not a voodoo word, but I find is very irritating that everyone goes on to talk about how expensive Azure and bing data is (when they don't know the slightest thing about it) and then go onto saying there will have to be a subscription model lol.  Yes it's expensive, but so is a traditional data center for multiplayer games - how are so many tripple A game titles running their servers without subscriptions?.  Azure and bing data is not exclusively used by MSFS, when Microsoft invests in these projects, they get their money back through loads of other projects that use those services.  The fact that MS owns them, means they can provide it to themselves at the lowest cost price.

3 hours ago, MattNischan said:

The flight simulator market is not enormous. It is big, and clearly MS wants to dip into casual players as well, but FSX, released in '06, is reported to have moved only 1M units by '08. Contrast that with a game like GTA V, which moved 110M units.

You mention GTA V, did you know at the time of it's release it was actually one of the most expensive games to develop in the world, coming in at over $265 million.  They made their money back very quickly through their large user base, but that did not stop them from milking it further.  They have multiplayer servers that they use to sell add-ons, new vehicles, ingame cash credit cards, weapons, houses, businesses, gamemodes - all of this has kept people playing on their servers and purchasing more and more items.  This means that some of their users would have spent over and above what the original game cost on just add-ons.  So how is GTA V really making money? It's through the add-ons, but on top of that they have users buying an insane amount of copies.  How many of us here have spent more money on add-ons than we have on the actual game (i.e. FSX) - I certainly have, way more.  I'll leave this here: https://gamerant.com/ubisoft-microtransaction-sales/ - there are plenty of logical ways for MS to make a good profit from this project.  My vote would be for them to develop add-ons themselves and sell them, open up a marketplace and charge a listing fee for "official" developers to sell their add-ons too.  That way they reuse what is already a proven model in this community.

3 hours ago, MattNischan said:

Furthermore, we also know that Asobo and MS both plan to make MSFS a platform that they will iterate on. Which means ongoing development costs as well.

Which can be offset by selling their additional work, i.e. though add-ons.

3 hours ago, MattNischan said:

What can one read from the tea leaves here? That MS is likely trying to decide if this project is more of a technical investment or more of a traditional game project. If they decide it's more of a traditional game project, I think there's no way to avoid a subscription-like model, whether that means that future periodic updates are paid or it's a true monthly subscription, because otherwise the number of units a game like this moves just doesn't turn a profit at that level of expenditure.

That's just plain wrong, a subscription model is not only easily avoidable but also something to be avoided.  If they go the subscription route, that will be suicide for this game.  Remember, flight simulators are boring and have a very limited user base (so not a lot will buy subscriptions) - nerds like me are happy to spend 18 hours flying from London to Australlia, but even I get bored after a while and put my simulator down for long periods at a time (times where no subscription will be paid, but Microsoft will still have to keep the lights on).  What brings me back? Normally a good new aircraft add-on.

EDIT: Also, just to add, the company that I work for as a Software Engineer uses a lot of Microsoft Azure services, as do we Amazon AWS and Google Cloud Platform, we pay all of these providers millions of pounds every month to host our services on them.  You think they ain't making money because someone is trying to run their tiny flight simulator program on them? Haha - these services are designed to serve tens of thousands of applications and companies.

Edited by KillerKlient
  • Like 1

1_logo2small.png

PVA1103 Robert Stevens

Share this post


Link to post
Share on other sites
11 minutes ago, Krakin said:

You do...not...know...what...you're talking about. You do not know how much Azure utilization is costing MS so stop talking like you do. You are completely ignoring what Phil Spenser has said.

Azure is an enormous department that is certainly not at all free of cost. And the video provided talks nothing of the Xbox division's costs at all, only that there was some available datacenter space for the streaming project. Actually, the way I take that is that Xbox's streaming platform is not running on top of the Azure platform at all, it just happens to reside in the same real estate.

It's not hard to make educated guesses here that are well within the order of magnitude, if even possibly within percentage points. The last yearly fiscal report from MS has the Intelligent Cloud division (which is almost solely Azure, but some other bits and bobs) making $38.9B in revenue and $13.9B in operating costs. Using that metric we can assume that end-user Azure costs have a profit margin built into them of about 35%, which is a decent amount, but not, say, 300%. Azure costs are not fixed for Microsoft; as additional subscribers add demand to the platform, they expand capacity with hardware, so while there is some economies of scale there, it's mostly on the software end which is definitely paid for, so I would expect that percentage to stay relatively stable.

Using that, let's focus on just one single component of the whole equation, bandwidth costs. In Azure, bandwidth costs are divided into global zones, with some being very cheap and some being quite expensive. Let's just be generous and assume most folks fall into a US or EU cost range, which would fall to about $0.0030/GB. Taking away the 35% means it probably costs MS about $0.00195/GB.

Let's also assume, on average, that you have about 40K concurrent users. Sometimes there will be more, sometimes less, but that's an average. Not a huge online game load, certainly (about the average-ish of EVE Online) but it's a niche game. Let's also say that scenery streaming comes it at just 10Mbps required (less than Netflix 4K at 25Mbps).

  • 40,000 * 10Mbps = 400,000Mbps = 400Gbps
  • 400Gbps / 8bits = 50GB/s
  • 50 * 60s * 60m * 24h * 30d = 129,600,000GB/mo
  • 129,600,000GB * $0.00195 = $252,720

A single 10GbE internet connection usually comes in at around $5-$10K/mo, even with a great contract, and just to serve the players you need 40x of those. ~$250K/mo for just bandwidth for a small game like this is not even close to out of the realm of possibility.

That cost is before we're talking about any storage, application servers, database servers, web apps, availability groups, etc. etc.

-Matt N


Matt Nischan
Architect | Lead Developer

ZF67cij.png

Share this post


Link to post
Share on other sites
On 10/29/2019 at 3:21 PM, MattNischan said:

Azure is an enormous department that is certainly not at all free of cost. And the video provided talks nothing of the Xbox division's costs at all, only that there was some available datacenter space for the streaming project. Actually, the way I take that is that Xbox's streaming platform is not running on top of the Azure platform at all, it just happens to reside in the same real estate.

I

-Matt N

You're assuming Microsoft charge the same amount for Azure to their low end external custmoers as they do internally to themselves? Microsoft have special rates for big customers and even better rates for themselves.

40K concurrent users in flight sim? Doubtful lol, there are tripple A titles out there that just about manage that.


1_logo2small.png

PVA1103 Robert Stevens

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

  • Tom Allensworth,
    Founder of AVSIM Online


  • Flight Simulation's Premier Resource!

    AVSIM is a free service to the flight simulation community. AVSIM is staffed completely by volunteers and all funds donated to AVSIM go directly back to supporting the community. Your donation here helps to pay our bandwidth costs, emergency funding, and other general costs that crop up from time to time. Thank you for your support!

    Click here for more information and to see all donations year to date.
  • Donation Goals

    AVSIM's 2020 Fundraising Goal

    Donate to our annual general fundraising goal. This donation keeps our doors open and providing you service 24 x 7 x 365. Your donation here helps to pay our bandwidth costs, emergency funding, and other general costs that crop up from time to time. We reset this goal every new year for the following year's goal.


    45%
    $11,465.00 of $25,000.00 Donate Now
×
×
  • Create New...