'Intel Arc B770 graphics card coming after all, release planned for later this year'

[-]

hardware-ModTeam@reddit

Thank you for your submission! Unfortunately, your submission has been removed for the following reason:

It is a submission that is largely speculative and/or lacks sufficient information to be discussed.

Rumours or other claims/information not directly from official sources must have evidence to support them. Any rumor or claim that is just a statement from an unknown source containing no supporting evidence will be removed.

Please read the the subreddit rules before continuing to post. If you have any questions, please feel free to message the mods.

[-]

TheArtBellStalker@reddit

If this was true (sounds more like huffing copeium to me) I would see this as a bad thing. The idea was to skip the B770 and focus on Celestial.

If we get this toward the end of the year, that means Celestial is lot longer off than we were hoping for.

[-]

Johnny_Oro@reddit

If Celestial dGPU is really on 18A-P, it's gonna take a while. Raichu did hint its going to be based on "Xe3P", so that could mean its fabbed in 18AP, which is an early 2027 thing.

[-]

Exist50@reddit

could mean its fabbed in 18AP, which is an early 2027 thing

Might be in practice, but it's supposed to be a 2026 node. Besides, Gelsinger cancelled that Celestial, so who knows if they'll have a future dGPU at all. Earliest point now would probably be 2028.

[-]

6950@reddit

He cancelled the laptop dGPUs in favor of big GPUs.

[-]

Johnny_Oro@reddit

Source on gelsinger cancelled celestial?

[-]

Exist50@reddit

The closest you'll find publicly is his remarks about "focusing more in integrated graphics". As you can see, Intel doesn't publish public roadmaps anymore.

[-]

Johnny_Oro@reddit

Sounds like a kind of a flimsy evidence to me. Intel can halt and continue R&D anytime they want. And I don't think their iGPU and dGPU R&D teams are so separate that they can only have one at a time. Like this G31 chip, it was in a limbo and I remember many of the legit battlemage leakers like Jaykin believed it's nothing more than an engineering sample, yet here we are, speculating again. Also meteor lake desktop that was cancelled abruptly, we have engineering sample and everything. I think the dGPU Gelsinger referred to when he said "less need for discrete graphics" is the G31 even.

It looks to me Intel has no product roadmaps anymore because nothing is certain. Seeing the positive reception of B580 seems to motivate them to mass produce B770, and B770's market performance as well as 18A's success will also affect their future decision for Celestial dGPU.

[-]

Exist50@reddit

Sounds like a kind of a flimsy evidence to me

I'm not using that as proof to justify making the statement, just pointing you to as much public evidence as there is.

Intel can halt and continue R&D anytime they want

Not really. They've laid off a ton of people in graphics. Sounds like it's pretty much a skeleton crew. You can't just reverse those kind of decisions on a dime even if you want to, and again, Intel's promised investors billions upon billions in spending cuts.

I think the dGPU Gelsinger referred to when he said "less need for discrete graphics" is the G31 even.

Celestial (as previously defined) was cancelled before any decision was made on G31. But they had the die pretty much finished and ready for tape out. Reviving a product that was still deep in execution is much harder, especially when the team is gone.

[-]

e-___@reddit

I wouldn't think so, they're playing catch-up with NVIDIA and AMD, falling behind Rubin and UDNA would not do them well

[-]

Exist50@reddit

Frequent launches are very expensive. Intel's trying to spend as little money as possible. They've claimed multiple rounds of billion-dollar spending cuts. Got to come from somewhere.

[-]

6950@reddit

They are going to launch at some public events lol

[-]

greiton@reddit

I think Intel GPU sales are much higher than predicted with that original strategy, and now they are in a position where launches are incredibly profitable, and they will do both.

[-]

mockingbird-@reddit (OP)

now they are in a position where launches are incredibly profitable

LOL, no

[-]

Exist50@reddit

and now they are in a position where launches are incredibly profitable

I think that's...optimistic to say the least. And remember, Intel's been cutting profitable parts of the business, so even if they've somehow dug out of their hole, that may still not be enough to save it.

[-]

greiton@reddit

they have been unable to keep them in stock. literally every unit produced get bought immediately and resold a 20%+ of MSRP.

[-]

Exist50@reddit

You're assuming they're producing many to begin with. And the situation isn't going to get better with 9060 and 5060 availability.

[-]

e-___@reddit

Still, if they want to be competitive on this space, they'll have to eat the losses while their market share starts consolidating, otherwise Arc will never make it past 2%

[-]

Exist50@reddit

Can they afford to "eat the losses"? They may instead prefer not to compete at all. Or at minimum, take the risk of a very long gap.

[-]

Winter_2017@reddit

I don't think Intel anticipated the endless demand for battlemage at $50+ over MSRP.

[-]

chefchef97@reddit

I like to imagine a bunch of rational people standing around at Intel speaking along the lines of "If we want to break into this market we're going to have to make some tough decisions, it's going to be a long time until we can price these- wait what the fuck"

[-]

ThankGodImBipolar@reddit

That’s fine if the additional time in the oven translates to a better launch and a more stable product (see RDNA 4). If Intel can improve the CPU overheard in their driver for Celestial (for example), then I’d consider it a win. Even if the B770 isn’t competitive, it will generate news cycles, and it’s going to be fast enough to be usable.

[-]

Exist50@reddit

it will generate news cycles, and it’s going to be fast enough to be usable

Generate news cycles for what? Reviews will compare it to Nvidia and AMD's latest and that's not going to paint a pretty picture. And the last thing Intel can afford to do is go through all the trouble of launching a GPU and get nothing but bad PR and wasted money.

[-]

ThankGodImBipolar@reddit

Intel releasing GPUs at all seems to generate positive press. As long as the cards aren’t fundamentally broken like launch Alchemist was, I think it’ll be a net positive. I’m just speculating though.

[-]

Not_Yet_Italian_1990@reddit

The issue is that B580 is already on the market. B770 will be a Battlemage card. There isn't really any "extra time in the oven." It's just a scaled-up B580.

Any improvements they make to CPU overhead would have been/would be accomplished by optimizing the B580, where it's likely more of an issue.

I guess maybe you can make the argument that they'll carry the lessons over to Celestial... but I don't know why a B770 release is necessary to do that.

To be completely clear, though, I do hope we get a B770, just to shake things up a bit. I just don't think that particular argument holds water. And, it's also possible that the B770 is a complete failure that kills Intel's GPU line altogether, so there's also some risk.

[-]

ThankGodImBipolar@reddit

I was talking about Celestial dGPUs getting extra time in the oven, not the B770. And you’re right, the B770 doesn’t need to come out for that to happen, but I would take a B770 release to mean that Celestial is at least a year out. That alone would make it a win in my eyes; I’m in no rush to see Celestial.

I don’t think it’ll be a complete failure either. It will probably be obviously uncompetitive when you compare its die size and perf/w to the other players, but it’ll probably work as well as the B580 (fine AFAIK), and it’ll definitely be faster. I doubt the sales of a potential B770 will justify finishing that die, but the sales of a 32GB Arc Pro card (or 64GB dual GPU card…) might.

[-]

Not_Yet_Italian_1990@reddit

It will probably be obviously uncompetitive when you compare its die size and perf/w to the other players

Maybe. But isn't Intel also using TSMC at this point?

In theory, they could drop a card early next year using TSMC 3nm that is a 5070/5070 Super killer, no?

It would probably be expensive, but it would be a year before Nvidia could do anything and they could use that time to buy up market share. And TSMC 3nm would be 2 1/2 years old at that point...

[-]

MiloIsTheBest@reddit

Even if the B770 isn’t competitive, it will generate news cycles

I'm all for the value of generating news cycles but boy oh boy would that be an expensive way to generate news cycles.

[-]

Dangerman1337@reddit

Maybe Celestial is 1H of 2027 on 18A-P since that is HVM for Q4 2026 (NVL Desktop seems to be on TSMC N2P for that reason mostly).

[-]

Exist50@reddit

18A-P should be a proper 2026 node. NVL's N2 usage has nothing to do with 18A's schedule. Can't ship without it regardless.

18A Celestial was cancelled, so assuming it's too late to revive it, we're looking at probably 2028 at best for what would have been called Druid.

[-]

Rollingplasma4@reddit

I want off this wild ride Intel confirm or deny it already.

[-]

mockingbird-@reddit (OP)

Intel confirms that it will release something sometime in the future.

[-]

wintrmt3@reddit

Source?

[-]

bubblesort33@reddit

That was before they released the workstation 48gb cards I thought. I assumed that was the announcement.

[-]

Tiny-Breadfruit4455@reddit

I'd imagine new CEO changing directions.

[-]

faverodefavero@reddit

B770 at the end of this year, and Celestial lineup at the end of 2026 trough very, very, early 2027. Then Druid at the end of 2027 trough very early 2028.

[-]

Exist50@reddit

Intel's not doing a yearly GPU cadence, period.

[-]

faverodefavero@reddit

They should, at least for now. To stay competitive and on people's, buyers, mind. That would be the perfect plan. Afterwards, they can (and should) slow down.

[-]

Exist50@reddit

They don't have the budget or the team to make that happen, even if there was a will. At this point they're walking a tightrope of having a future in client dGPU at all.

[-]

kingwhocares@reddit

They were probably waiting to see RTX 5060 and RX 9060XT performance before releasing and position it accordingly.

[-]

Orelha3@reddit

Eh, they should be focusing on fixing their driver overhead shenanigans, else the B770 will only be viable on a 9800x3D.

[-]

Front_Expression_367@reddit

technically speaking, it is more likely that people pairs stronger CPUs for a higher-tiered GPU. It still won't look pretty, but Intel's pricing on the GPU will still be a bigger factor imo.

[-]

kingwhocares@reddit

If it is a long time process to fix driver issues, it's best to not hold everything down just for one thing.

[-]

Exist50@reddit

It's better to launch first and adjust pricing after than to wait till your competition already got those sales.

[-]

kingwhocares@reddit

Given AMD's supply, doubt. And people buy them even when new gens are released.

[-]

Exist50@reddit

Given AMD's supply, doubt

AMD's supply is going to still be way higher than Intel's. And any sale lost is a sale lost.

And people buy them even when new gens are released.

Some still might, but people aren't going to sell their 9060 for a B770. This isn't some intentional strategy.

[-]

kingwhocares@reddit

AMD's supply is going to still be way higher than Intel's. And any sale lost is a sale lost.

Their supply will be lower than demand and prices will be high and not reflecting the price set by AMD.

[-]

Exist50@reddit

Maybe they will, maybe then won't. By the time B770 could come out, it would have normalized somewhat if there's even early gouging to begin with. Regardless, if prices are inflated because of demand, that implies a lot of people are still buying them. Those are then people de facto out of the market for B770.

[-]

mockingbird-@reddit (OP)

Stay tuned!!