Xbox Business Update Podcast | Xbox Everywhere Direction Discussion

[Tkumpathenurpahl]

Xbox is now a mature brand so I'm not sure people are still interested in seeing what they have to market considering they trashed all of their notably franchises (Halo, Gears, Forza), and even if you do like those games they are bringing them to PlayStation/PC/your internet browser via XCloud so why bother buying their hardware?

All good points, but I think they can still make their hardware enticing.

1) launch a portable which can run Series S games. And a home console which is both powerful and easy to extract performance from i.e. no split bandwidth memory.

2) dual boot for both devices. Windows and Xbox OS. Xbox OS being more suited for portability or TV screens, as well as a smoother update process, notifications etc.

3) stop day and date launches for PC. Focus on the consoles first, and port to PC in 6-12 months. Anyone who buys on the Xbox console store gets the PC Xbox store version for free. FOMO has been cultivated across pretty much every culture on earth - they should capitalise on that while making life easier for their enormous amount of developers.

4) name them sensibly. "Xbox 5 Portable" and "Xbox 5 Home" off the top of my head are easy for the public to grasp.

 

[XboxKING]

If they deliver great new AAA titles this thing can survive otherwise it could end up as a dud. Very experimental territory Phil has pushed aggressively into.

That is what I said… back in 2017

 

[Charlietus]

All good points, but I think they can still make their hardware enticing.

1) launch a portable which can run Series S games. And a home console which is both powerful and easy to extract performance from i.e. no split bandwidth memory.

2) dual boot for both devices. Windows and Xbox OS. Xbox OS being more suited for portability or TV screens, as well as a smoother update process, notifications etc.

3) stop day and date launches for PC. Focus on the consoles first, and port to PC in 6-12 months. Anyone who buys on the Xbox console store gets the PC Xbox store version for free. FOMO has been cultivated across pretty much every culture on earth - they should capitalise on that while making life easier for their enormous amount of developers.

4) name them sensibly. "Xbox 5 Portable" and "Xbox 5 Home" off the top of my head are easy for the public to grasp.

Is it even possible to launch a series s portable at mass market prices in 2026? Steam deck, ROG ally and the others are pretty far from matching a series s, so you would need something like a 2nm chip with lpddr6 memory. That would be 600$/€ in the best case scenario, 700$/€ would be more realistic.

Launching games 12 months later on PC... It's not possible. Aside from Phil and (most of all) Nadella always saying that they want to reach more users on other platforms, you can't sustain big games just on the Xbox. With how many studios Microsoft has, it's barely sustainable with Xbox and PC, as we have seen in recent times.

 

[Shifty Geezer]

That is what I said… back in 2017

You joined May 2024. Was that under a different account here?

 

[Tkumpathenurpahl]

Is it even possible to launch a series s portable at mass market prices in 2026? Steam deck, ROG ally and the others are pretty far from matching a series s, so you would need something like a 2nm chip with lpddr6 memory. That would be 600$/€ in the best case scenario, 700$/€ would be more realistic.

The Series S consumes 74W during gaming. 5nm would take that to 51.8W and N4P would take it to 40.4W. So a Nintendo Switch type of portable wouldn't be viable, but a laptop type would, with scope for Switch size once at 3nm or 2nm.

It would require 12Gbps LPDDR6 on a 192-bit bus if this table is correct.

So I certainly agree that it'd be fairly pricey, but inclusive of Windows, I think many people would find themselves able to justify the expense.

Launching games 12 months later on PC... It's not possible. Aside from Phil and (most of all) Nadella always saying that they want to reach more users on other platforms, you can't sustain big games just on the Xbox. With how many studios Microsoft has, it's barely sustainable with Xbox and PC, as we have seen in recent times.

I agree that this is pretty unlikely, given the commitments made by the bigwigs, I just think they're wrong on this matter.

Halo Infinite showed us the problem of developing simultaneously for a bunch of hardware configurations. I say free the developers of art from a lot of technical constraints by allowing them to focus exclusively on a high end and low end version of their game, and then hand it over to an exclusively technical porting team to get it running on infinite PC configurations and other consoles.

Gamers would have reason to buy Xbox consoles, and the technical porting team could have close ties to the Direct X team to help identify issues with scaling across performance tiers.

 

[Charlietus]

The Series S consumes 74W during gaming. 5nm would take that to 51.8W and N4P would take it to 40.4W. So a Nintendo Switch type of portable wouldn't be viable, but a laptop type would, with scope for Switch size once at 3nm or 2nm.

It would require 12Gbps LPDDR6 on a 192-bit bus if this table is correct.

So I certainly agree that it'd be fairly pricey, but inclusive of Windows, I think many people would find themselves able to justify the expense.



I agree that this is pretty unlikely, given the commitments made by the bigwigs, I just think they're wrong on this matter.

Halo Infinite showed us the problem of developing simultaneously for a bunch of hardware configurations. I say free the developers of art from a lot of technical constraints by allowing them to focus exclusively on a high end and low end version of their game, and then hand it over to an exclusively technical porting team to get it running on infinite PC configurations and other consoles.

Gamers would have reason to buy Xbox consoles, and the technical porting team could have close ties to the Direct X team to help identify issues with scaling across performance tiers.

Honestly, what I would do is start a huge marketing campaign that puts eyes on the brand. A 400$ portable that goes with it and then pray that some of their games get universal appraise.

But the switch 2 will exist soon, and that is going to be a black hole.

Ps: yeah, no, I don't know what they can actually do
I wouldn't want to be in that situation

 

[XboxKING]

You joined May 2024. Was that under a different account here?

I said it IRL after MS themselves stated they would become the Netflix of gaming; I said they would need better games compared to what they released that generation as the occasional cuphead or mid-tier Forza would not cut it
I don’t have a different account here

 

[Tkumpathenurpahl]

Honestly, what I would do is start a huge marketing campaign that puts eyes on the brand. A 400$ portable that goes with it and then pray that some of their games get universal appraise.

But the switch 2 will exist soon, and that is going to be a black hole.

Ps: yeah, no, I don't know what they can actually do
I wouldn't want to be in that situation

The trouble there is that they'd be fighting Nintendo on Nintendo's terms, which are strongly favourable towards Nintendo.

So yeah, I agree that I wouldn't want to be in that situation. But it's situations like this that warrant something of a gamble. And just as Nintendo realised great success by merging their portable and home consoles, MS may do so by merging their consoles and Windows.

 

[eastmen]

The Series S consumes 74W during gaming. 5nm would take that to 51.8W and N4P would take it to 40.4W. So a Nintendo Switch type of portable wouldn't be viable, but a laptop type would, with scope for Switch size once at 3nm or 2nm.

It would require 12Gbps LPDDR6 on a 192-bit bus if this table is correct.

So I certainly agree that it'd be fairly pricey, but inclusive of Windows, I think many people would find themselves able to justify the expense.



I agree that this is pretty unlikely, given the commitments made by the bigwigs, I just think they're wrong on this matter.

Halo Infinite showed us the problem of developing simultaneously for a bunch of hardware configurations. I say free the developers of art from a lot of technical constraints by allowing them to focus exclusively on a high end and low end version of their game, and then hand it over to an exclusively technical porting team to get it running on infinite PC configurations and other consoles.

Gamers would have reason to buy Xbox consoles, and the technical porting team could have close ties to the Direct X team to help identify issues with scaling across performance tiers.

you are assuming they will just use zen2/rdna 2. Moving to more power efficient tech like zen5/6 and rdna 4 can get power demands much further

halo infinite single player was well received. The multiplayer had its own issues

 

[rntongo]

https://twitter.com/x/status/1823859033880846782

If Indeed Forza is coming to PS5 this year then Gears as well will be coming to the PS5. MS seems to be making a business decision to boost software revenues considering how badly they managed their own hw. I think this is good move in the short to medium term

 

[Tkumpathenurpahl]

you are assuming they will just use zen2/rdna 2. Moving to more power efficient tech like zen5/6 and rdna 4 can get power demands much further

True. Given Microsoft's skills with abstraction, it'll be interesting to see what minimum hardware will be necessary to play Series S games. Fingers crossed such a portable (whether tablet or laptop) materialises.

halo infinite single player was well received. The multiplayer had its own issues

I've only heard positive things about its gameplay, but it was delayed by a year, was fairly graphically underwhelming, and IIRC suffered a number of technical issues. I don't think most of those (particularly the first two) would have been the case if 343 only had to focus on Series S/X versions.

 

[eastmen]

True. Given Microsoft's skills with abstraction, it'll be interesting to see what minimum hardware will be necessary to play Series S games. Fingers crossed such a portable (whether tablet or laptop) materialises.

Looking at performance its really rdna 2/3 and bandwidth that is the issue. An 8 core zen 5 should be more than fast enough at lower speeds

the series s has
zen 2 @3.4/3.6 with the ipc improvements from zen 2 through zen 5 (or even zen 6 by that time) should allow for lower speeds at lower wattage.

Rdna 2 20cu @1.56ghz we are pretty close with the rdna 3.5 apu but rdna 4 may bring enough performance gains that you can get away with less cu's at lower clocks and have the same performance. We wont know until later this year or early next year

Bandwidth is where I see the issues 224GB/s is going to be hard to recreate in an affordable handheld I think.


if they figure out bandwidth and go with rdna 4 they could actually have features turned off in the s version turned back on for a mobile

I've only heard positive things about its gameplay, but it was delayed by a year, was fairly graphically underwhelming, and IIRC suffered a number of technical issues. I don't think most of those (particularly the first two) would have been the case if 343 only had to focus on Series S/X versions.

The game was great and I thought it looked great. The multiplayer was the issue for me.

 

[Cappuccino]

Halo Infinite didn’t really do that well. The multiplayer was a let down and the campaign was probably the most forgettable of the series. Graphically it looked like an Xbox One game… which it was. I thought Halo 5 looked better at times.

‘Craig’ was a meme for a time for a reason. They had to delay the game because the first trailer looked so bad lol. The final product didn’t look all that much better.

 

[Charlietus]

Halo Infinite didn’t really do that well. The multiplayer was a let down and the campaign was probably the most forgettable of the series. Graphically it looked like an Xbox One game… which it was. I thought Halo 5 looked better at times.

‘Craig’ was a meme for a time for a reason. They had to delay the game because the first trailer looked so bad lol. The final product didn’t look all that much better.

Infinite situation was interesting. The campaign was praised at first, but now the sentiment is more negative (I didn't like it from the beginning, only two biomes? That's just poor compared to the other games). The multiplayer instead was controversial at launch with not much content, but now it has gotten better.

If you judge it as THE Xbox game, that's really poor.

 

[see colon]

Looking at performance its really rdna 2/3 and bandwidth that is the issue. An 8 core zen 5 should be more than fast enough at lower speeds

the series s has
zen 2 @3.4/3.6 with the ipc improvements from zen 2 through zen 5 (or even zen 6 by that time) should allow for lower speeds at lower wattage.

Rdna 2 20cu @1.56ghz we are pretty close with the rdna 3.5 apu but rdna 4 may bring enough performance gains that you can get away with less cu's at lower clocks and have the same performance. We wont know until later this year or early next year

Bandwidth is where I see the issues 224GB/s is going to be hard to recreate in an affordable handheld I think.


if they figure out bandwidth and go with rdna 4 they could actually have features turned off in the s version turned back on for a mobile

The game was great and I thought it looked great. The multiplayer was the issue for me.

We've seen how they can adjust the resolution and settings like AF without changing game code, so it would be interesting to see what they could do for a lower power device. Theoretically, you would need less bandwidth if you lowered the resolution. I'd also be curious to see if they revert back to having some embedded memory. If you know what your output resolution is like you would on a handheld, you could have a small pool of super fast memory for your framebuffer, which would allow you to have a larger pool of less performant memory.

Halo Infinite didn’t really do that well. The multiplayer was a let down and the campaign was probably the most forgettable of the series. Graphically it looked like an Xbox One game… which it was. I thought Halo 5 looked better at times.

‘Craig’ was a meme for a time for a reason. They had to delay the game because the first trailer looked so bad lol. The final product didn’t look all that much better.

I always though Halo Infinite looked fine. Maybe not as cutting edge as the first few games in the series, but very much like a high resolution open world Halo game. Which is what it was.

 

[Tkumpathenurpahl]

the series s has
zen 2 @3.4/3.6 with the ipc improvements from zen 2 through zen 5 (or even zen 6 by that time) should allow for lower speeds at lower wattage.

Rdna 2 20cu @1.56ghz we are pretty close with the rdna 3.5 apu but rdna 4 may bring enough performance gains that you can get away with less cu's at lower clocks and have the same performance. We wont know until later this year or early next year

Potentially, but that's very much hypothetical. We've yet to see anything from Microsoft where they run the exact same code on lesser hardware: the X360 games running on the XB1 required each game to be downloaded in a recompiled form which included its own emulator; the Series S runs XB1 versions, rather than XB1X.

So I'd expect an XB Portable to have greater performance than the Series S if we're looking at native games utilising a newer architecture, but I'm yet to be convinced that fewer CU's at a lower clockspeed would run Series S games natively.

Bandwidth is where I see the issues 224GB/s is going to be hard to recreate in an affordable handheld I think.

I agree if we're dealing with an affordable handheld, but I don't think trying to fight Nintendo directly and cheaply is wise.

10.667Gbps LPDDR6 on a 192 bit bus would provide 228GB/s bandwidth using 8 chips of memory. Make 6 of them 2GB and 2 of them 4GB and developers would have 16GB of memory at a full 228GB/s while the OS would have 4GB at 57GB/s.

That would:
- be ample to outperform the Switch 2 the entire generation
- sufficient to trade blows with the PS5
- keep life simple for developers because they don't have to worry about different pools of bandwidth
- come down in cost over time because LPDDR6 won't be expensive forever and 10.667Gbps is the cheapest one
- be desirable in spite of being more expensive than the Switch 2 if the device can dual boot into Windows

I really think it would be as sensible of a move as the PS4 going with 8GB of GDDR5. It was an expense, but it absolutely paid for itself.

if they figure out bandwidth and go with rdna 4 they could actually have features turned off in the s version turned back on for a mobile

Steady on now, let's not get greedy. You can't have your cake and eat it.

The game was great and I thought it looked great. The multiplayer was the issue for me

Going by what I've seen, it looked good, but not at the level it deserved to be. It looked really good for something running on infinite PC configurations and four Xbox consoles, but it could've looked a good deal better if only the Series S/X were the focus for launch.

Kind of like we saw with Horizon and God of War. Both were pretty, but they were hamstrung by having to run on the base PS4, and developer time had to be spent getting them running on 3 devices. Not to mention the slipping through crevice loading screens (-_-)

I'd also be curious to see if they revert back to having some embedded memory

Have we seen any laptop RDNA3 APU's using Infinity Cache? I'm doubtful of EDRAM type embedded memory (and I'm doubtful that either embedded memory or IC could make lower than Series S bandwidth sufficient for Series S games) but I think some amount of IC could help a portable punch above its weight for native titles.

 

[QPlayer]

Mobilizing the Series S wouldn't make much sense in my opinion. Future games will require much more modern hardware. Therefore, if MS plans a mobile Xbox, it will probably be a new architecture that includes an NPU for local AI.
And in terms of power consumption, modern architecture is also required.

Due to the small mobile screen, in terms of performance, it is enough to aim for a lower render resolution with some kind of image enhancement. Therefore, a 10 TFlops GPU can be suitable, which in mobile mode renders around 600p / 20 Watts, in docked mode it can come with a resolution of 1080p / 60 Watts.

 

[rntongo]

Mobilizing the Series S wouldn't make much sense in my opinion. Future games will require much more modern hardware. Therefore, if MS plans a mobile Xbox, it will probably be a new architecture that includes an NPU for local AI.
And in terms of power consumption, modern architecture is also required.

Due to the small mobile screen, in terms of performance, it is enough to aim for a lower render resolution with some kind of image enhancement. Therefore, a 10 TFlops GPU can be suitable, which in mobile mode renders around 600p / 20 Watts, in docked mode it can come with a resolution of 1080p / 60 Watts.

With the amount of money Xbox has spent on ARM, I think a mobile device should use ARM for the CPU. If I'm not mistaken they had plans to move the next gen Xbox possibly to ARM as well. It depends on what feedback they get back from devs as well as the technical teams at AMD and the MS hw gaming team. But for me it makes sense to make the change for a mobile device. Give it a separate SKU. Otherwise even Steam OS is going after Windows as a gaming platform since its going to be available on other devices.

 

[Tkumpathenurpahl]

Mobilizing the Series S wouldn't make much sense in my opinion. Future games will require much more modern hardware. Therefore, if MS plans a mobile Xbox, it will probably be a new architecture that includes an NPU for local AI.
And in terms of power consumption, modern architecture is also required.

Due to the small mobile screen, in terms of performance, it is enough to aim for a lower render resolution with some kind of image enhancement. Therefore, a 10 TFlops GPU can be suitable, which in mobile mode renders around 600p / 20 Watts, in docked mode it can come with a resolution of 1080p / 60 Watts.

Agreed on all points. I do, however, think that taking Series S power consumption and looking at how that changes across nodes gives us a rough idea of the power consumption required for it to run Series S games without them being recompiled.

It'll be interesting to see how newer versions of Zen and RDNA lower relative power consumption for matched/exceeded Series S specs when running Series S games b/c and how much more performant native games are.

I think Strix Halo can boost up to 3Ghz, so 20*128*3=7.68TF or 15.36TF dual-SIMD. It would be a bit bandwidth starved at 228GB/s but IC would help it punch above its weight there.

Pretty good for a docked mode IMO.

 

[Nisaaru]

I don't even get the idea of wasting so much HW on trying to play these games on mobile. It's not like mobile gamers truly care for quality due display and interface limits so Cloud should be good enough for 99% of their needs and you could even play longer due the low energy footprint.

But then gaming was never really rational anyway.