MSFT CY24Q4 Results are in

Echohead2

Ars Legatus Legionis
61,027
though I’m still not sold on the long-term viability of GamePass as a primary business model) and limited game selection compared to the Playstation.
Well...it could be very viable. I mean $15/month is $180/year/subscriber. How many subscribers do you need before it is viable? 10mil would be $1.8B/year. 100mil would be $18B/y. I suspect that 1mil isn't enough and 100mil would definitely be viable. So somewhere between 1 and 100mil is the tipping point. At lest that is how I see it.
 

Louis XVI

Ars Legatus Legionis
11,138
Subscriptor
Well...it could be very viable. I mean $15/month is $180/year/subscriber. How many subscribers do you need before it is viable? 10mil would be $1.8B/year. 100mil would be $18B/y. I suspect that 1mil isn't enough and 100mil would definitely be viable. So somewhere between 1 and 100mil is the tipping point. At lest that is how I see it.
Isn’t the answer going to depend on how much it costs to get that revenue? If you’re bringing in $18B/y, but it costs $25B/y, that doesn’t strike me as particularly viable, but if you’re spending $9B/y to make $18B/y, you’re doing great. Microsoft has been spending crazy money to acquire studios to generate product to feed the beast, and there certainly seem to be rumblings from MS Powers That Be that the way things have been going, revenues don’t justify the expense.
 

Echohead2

Ars Legatus Legionis
61,027
Isn’t the answer going to depend on how much it costs to get that revenue? If you’re bringing in $18B/y, but it costs $25B/y, that doesn’t strike me as particularly viable, but if you’re spending $9B/y to make $18B/y, you’re doing great. Microsoft has been spending crazy money to acquire studios to generate product to feed the beast, and there certainly seem to be rumblings from MS Powers That Be that the way things have been going, revenues don’t justify the expense.
Right, but you can't count the cost of the acq in the yearly cost. How much do they spend developing all of the games. Activision was spending about $6B/Year. Given its size, I doubt the rest of MS's developers are even that much. So $9B is probably a tad high, but close enough. So we both agree that 100mil subcribers would be good. Looks like around 50mil would be breaking even (well...you also have stand alone game sales from that cost, so probably lower). Last numbers I found were 34mil subscribers. That was before acquisitions so, hopefully they will be driving subscribers.
 

Louis XVI

Ars Legatus Legionis
11,138
Subscriptor
Right, but you can't count the cost of the acq in the yearly cost. How much do they spend developing all of the games. Activision was spending about $6B/Year. Given its size, I doubt the rest of MS's developers are even that much. So $9B is probably a tad high, but close enough. So we both agree that 100mil subcribers would be good. Looks like around 50mil would be breaking even (well...you also have stand alone game sales from that cost, so probably lower). Last numbers I found were 34mil subscribers. That was before acquisitions so, hopefully they will be driving subscribers.
Why can’t you count the cost of the acquisitions? The PTB at MS who are breathing down Spencer’s neck to get profitable sure seem to be concerned about all the costs, including the acquisitions.
 

Echohead2

Ars Legatus Legionis
61,027
Why can’t you count the cost of the acquisitions? The PTB at MS who are breathing down Spencer’s neck to get profitable sure seem to be concerned about all the costs, including the acquisitions.
Well, you can, you just need to amortize them over like 20-30 years, maybe longer. I mean if you buy a house you don't say you lost $500k that year do you? Plus you still have the asset...just like them.
 

Nevarre

Ars Legatus Legionis
24,372
Well, you can, you just need to amortize them over like 20-30 years, maybe longer. I mean if you buy a house you don't say you lost $500k that year do you? Plus you still have the asset...just like them.

The acquisition part is the not-straightforward part because it's not all first-party studios who don't have a choice in the matter on how they want to be compensated for joining your subscription.

Publishers need to weigh the fees paid by (in this case) Microsoft against the value of their IP. They don't want to be paid too little for something that might end up being a huge hit... except that half the 'sales' got siphoned off into a subscription deal they made on the cheap. Microsoft likewise doesn't want to pay top dollar for a game that promises the moon and the stars and ends up with <100 player counts a month after launch.

A lot of those titles will also be licensing content. We're seeing titles go dark because otherwise viable games are tied to a larger franchise and the publisher's contract with $movie/$TVshow/$mediaempire is up, or they licensed music for the in-game radio station, and that license is up and nobody wants to renegotiate or fix it in any other way. Boom, game dies for a stupid reason.

That's before ongoing support costs. Games are less patch-reliant but not zero patch-reliant. There's ongoing server maintenance overhead for online games-- sometimes very little, sometimes significant. Sometimes the old server infrastructure was set up by someone who left 10 years ago and nobody has the know how to rebuild them.

So you may pay to have access to an asset for a period of time, but it's not always exclusive access and that asset depreciates at a highly variable rate. Games usually have a pretty high roll-on, roll-off within the first month or so. Some settle into long-term stable high player counts, some dwindle to a small but still active player base, and some just dwindle down to almost nothing. If you're really unlucky that dwindling takes a few weeks not years...
 

theevilsharpie

Ars Scholae Palatinae
1,448
Subscriptor++
I'm just musing here but Microsoft building an ARM-based gaming device could help solve several significant challenges the company is currently facing. The first challenge is developing native support for Windows on ARM applications, specifically games. Creating an ARM-based gaming device would help jump-start this effort by engaging both internal game developers and third-party developers.

Building an ARM binary isn't difficult -- it's just a recompile for a different architecture. Developers have already had to deal with multi-arch development in the XBox 360 days when it was running on PowerPC and PCs were on x86, and continue to exercise that in modern times since PCs and XBox/PS are x86, and Phones/Apple devices/Nintendo Switch are ARM.

The reason game developers don't provide native ARM binaries for ARM on Windows is because the target market is practically non-existent, and supporting the platform would almost certainly cost more than the revenue it would bring in.

The second challenge is that the Xbox has struggled to distinguish itself from the Sony PlayStation and is often considered an inferior device. Designing a custom ARM processor could help position the Xbox as a unique and competitive platform.

There's nothing about a processor being ARM that inherently makes it superior to x86, and people who buy game consoles don't know or care about the underlying technical trivia like what instruction set it uses. It would also complicate things like backwards compatibility, since a hypothetical next-generation ARM XBox wouldn't be able to run past generation XBox software without using emulation or developers taking the time/effort to create an ARM binary for their past-gen games.

We've seen through Apple that ARM processors can provide industry-leading performance, and from the Nintendo Switch that an ARM processor can successfully power a gaming console. This device could be handheld or perhaps the next generation of gaming consoles.

No one doubts that ARM can power a game console. However, games are GPU limited, and GPU performance is an area where even the XBox Series S outperforms Apple's hardware until you reach the SKUs costing thousands of dollars.

There is no reason to believe that XBox moving to ARM will do anything other than waste Microsoft's money.
 

LordDaMan

Ars Legatus Legionis
10,760
Building an ARM binary isn't difficult -- it's just a recompile for a different architecture. Developers have already had to deal with multi-arch development in the XBox 360 days when it was running on PowerPC and PCs were on x86, and continue to exercise that in modern times since PCs and XBox/PS are x86, and Phones/Apple devices/Nintendo Switch are ARM.

For an unoptimized app yes, that is the case. For something like an Xbox game that may be a very bad thing as console games tend to be highly tailored to take full advantage of a particular hardware set and it's various quirks to eek out as much performance as possible.

It's why you can;t just do that with most OSes as they cinrain so much assembly and othe rlow level code that porting it takes a lot more work and/or a rewrite of some section

There's nothing about a processor being ARM that inherently makes it superior to x86, and people who buy game consoles don't know or care about the underlying technical trivia like what instruction set it uses. It would also complicate things like backwards compatibility, since a hypothetical next-generation ARM XBox wouldn't be able to run past generation XBox software without using emulation or developers taking the time/effort to create an ARM binary for their past-gen games.

The thing that in this particular case, the ARM based xbox would have to be far more powerful then current hardware to compensate for the emulation layer to run the old games. Have your old game running even slightly slower would be pretty much the end of the arm xbox
 

wrylachlan

Ars Legatus Legionis
13,683
Subscriptor
There's nothing about a processor being ARM that inherently makes it superior to x86
I’m not sure this is true. It’s extremely hard to tease out whether the ISA affords any actual technical advantage since it’s hard to find processors on the same process node that are targeting the same TDP - Lunar Lake should be interesting in this regard. But it isn’t unreasonable to assume that maintaining compatibility with a much older ISA would have some non-zero performance cost on X86.

But ignoring the technical side for a minute, the obvious benefit of ARM over X86 would be cost. Having a similar performance as your competitor at a lower cost is a big win in the console world.

Thinking through a putative next gen console SoC it seems pretty likely that CPU will be the least important component - with the GPU and ML processing blocks vastly more important. Microsoft could throw an off-the-shelf Cortex at the problem and be in a really good place.
 

Echohead2

Ars Legatus Legionis
61,027
Having a similar performance as your competitor at a lower cost is a big win in the console world.
This generation is kind of weird...4 years in and the prices are the same. That $500 console should be $300 by now or something. If the Series X was $300, I don't think you would be seeing sales dropping (at least based on units).
 

ant1pathy

Ars Tribunus Angusticlavius
6,731
No one doubts that ARM can power a game console. However, games are GPU limited, and GPU performance is an area where even the XBox Series S outperforms Apple's hardware until you reach the SKUs costing thousands of dollars.
Isn't this much more about the decision about where to allocate transistors for the device, rather than anything inherent in the ARM build itself? If the design goal was to spend 70% (90%?) of the transistor budget on GPU, you'd get very different results.
 
I’m not sure this is true. It’s extremely hard to tease out whether the ISA affords any actual technical advantage since it’s hard to find processors on the same process node that are targeting the same TDP - Lunar Lake should be interesting in this regard. But it isn’t unreasonable to assume that maintaining compatibility with a much older ISA would have some non-zero performance cost on X86.

But ignoring the technical side for a minute, the obvious benefit of ARM over X86 would be cost. Having a similar performance as your competitor at a lower cost is a big win in the console world.

Thinking through a putative next gen console SoC it seems pretty likely that CPU will be the least important component - with the GPU and ML processing blocks vastly more important. Microsoft could throw an off-the-shelf Cortex at the problem and be in a really good place.
ARM is not inherently cheaper. It's just more efficient than x86 in mobile packages. There is zero reason to assume that an ARM focused on a console that could compete on performance with Xbox and Playstation would be cheaper.
Certainly not from any of the companies that could currently build such a processor.
 

wrylachlan

Ars Legatus Legionis
13,683
Subscriptor
Isn't this much more about the decision about where to allocate transistors for the device, rather than anything inherent in the ARM build itself? If the design goal was to spend 70% (90%?) of the transistor budget on GPU, you'd get very different results.
I don’t think that’s it. Apple’s GPUs use a ton of transistors. The difference is that their design goal is “good performance at low power draw” while the console’s design goal is “great performance at… fuck it, electricity is cheap!”

Console GPUs run way, way hotter than an M series chip and are fundamentally designed to hit those exceptionally high and exceptionally hot frequencies. They also use incredibly fast (and expensive) memory that also uses dramatically more juice than Apple’s LPDDR. If you took a n Apple GPU, bolted on HBM and stuck it on a massive cooler so you could crank the frequency it would probably be a pretty competitive chip.
 
  • Like
Reactions: ant1pathy

LordDaMan

Ars Legatus Legionis
10,760
I don’t think that’s it. Apple’s GPUs use a ton of transistors. The difference is that their design goal is “good performance at low power draw” while the console’s design goal is “great performance at… fuck it, electricity is cheap!”

Console GPUs run way, way hotter than an M series chip and are fundamentally designed to hit those exceptionally high and exceptionally hot frequencies. They also use incredibly fast (and expensive) memory that also uses dramatically more juice than Apple’s LPDDR. If you took a n Apple GPU, bolted on HBM and stuck it on a massive cooler so you could crank the frequency it would probably be a pretty competitive chip.
There is more chance of Steve Jobs coming back from the dead and declaring windows the only one true OS and devoting his 2nd life to promoting windows and Microsoft in general then ever seeing a m series chip in a gaming console.

Alos the M3 isn't exactly the end all and be all of GPUs. It's roughly comparable to a mid-tier 3 year old RTX 3060. You know what else is about the same to somewhat more powerful than that? The current GPU in the xbox series x!.
 

Mark086

Ars Legatus Legionis
10,873
There is more chance of Steve Jobs coming back from the dead and declaring windows the only one true OS and devoting his 2nd life to promoting windows and Microsoft in general then ever seeing a m series chip in a gaming console.

Alos the M3 isn't exactly the end all and be all of GPUs. It's roughly comparable to a mid-tier 3 year old RTX 3060. You know what else is about the same to somewhat more powerful than that? The current GPU in the xbox series x!.
It's like you can read; but then didn't bother.
 

Nevarre

Ars Legatus Legionis
24,372
Why is nobody talking about the absolute horror that is Recall?

It's possible that it could come back, but at this point I think it's been "Microsoft BOB'ed."

The backlash was strong enough that they won't try that tactic again-- certainly not without radical changes to their approach and a more limited scope.
 
  • Like
Reactions: md5crypto
It's possible that it could come back, but at this point I think it's been "Microsoft BOB'ed."

The backlash was strong enough that they won't try that tactic again-- certainly not without radical changes to their approach and a more limited scope.
They better make it such that it requires a separate install.
 
  • Like
Reactions: jackwagon