InnerworldIntel announced plans to start making GPUs, challenging NVIDIA's dominance
At least they are admitting the Intel ARC was more of a joke rather than a graphics card.
Intel ARC is no joke. Technologically it’s very capable, they just never really scaled it to compete on any higher level…
AdrianTheFrogAt the datacenter scale Gaudi 3 was pretty good, at least when it came out.
WioumI had to check the date on the article. They’ve been making GPUs for 3 years now, but I guess this announcement–although weird–is a sign that Arc is here to stay, which is good news.
This article was based off what the CEO said at the Second Annual AI Summit, following the news of their new head of GPU hire who says he “will lead GPU engineering with a focus on AI at Intel”. The AI pivot is the actual news.
Just what every consumer needs. More AI focused chips.
Intel just trying to cash in on the AI hype to buy the sinking ship, as far as investors are concerned.
It feels like TechCrunch is allowing a drunk Ai to write all its articles now.
Weird, they’re a bit late boarding this train as it already starts to derail…MS just stumbled hard as their AI shit isn’t paying off and it drives consumers away.
Man watching the stock the past few days is just chefs kiss
It’s not even a pivot. They’ve been focusing on AI already. I’m sure they want it to seem like a pivot (and build up hype); the times before apparently just having the hardware and software wasn’t enough. nobody cared when the gaudi cards came out, nobody uses sycl or onednn, etc
Oh so they will actually not focus on GPUs as end consumer products for you and me. They’re just like Nvidia and AMD. This news really just shows how cooked gaming is.
I don’t know, perhaps gaming will get rejected AI chips with a few cores broken. The chip design requirements are slightly different but not completely foreign
Reygle
LeonThe actual chips are farmed out to TSMC, I don’t believe they’ve made any in house so I’m guessing maybe they’ve decided that they’re going to do that sometimes now? But then, even some of their CPUs are made by TSMC so I could be on a very wrong path.
TSMC is how they stay competitive; that’s what everyone else uses
Intel is still catching up with 18A
The 18A production node itself is designed to prove that Intel can not only create a compelling CPU architecture but also manufacture it internally on a technology node competitive with TSMC’s best offerings.
You are a bit out of date. I cant say what I know, but tsmc is just one player now. Semiconductor industry is about to make some jumps.
You design semiconductors for a living? Or just read articles?
They want to make Celestial on 18A, no?
DiplomjodlerWhat the fuck? What kind of idiotic article is that? Did Techcrunch go down the drain too?
Doesn’t Nvidia have $5bi stakes of intel? I wonder how that influences their decisions.
Well that article was a waste of space. Intel has already stepped into the GPU market with their ARC cards, so at the very least the article should contain a clarification on what the CEO meant.
And I see people shitting on the arc cards. The cards are not bad. Last time I checked the B580 had performance comparable to the 4060 for half the cost. The hardware is good, it’s simply meant for budget builds. And of course the drivers have been an issue, but drivers can be improved and last time I checked Intel is actually getting better with their drivers. It’s not perfect but we can’t expect perfect. Even the gold standard of drivers, Nvidia, has been slipping in the last year.
All is to say, I don’t understand the hate. Do we not want competition in the GPU space? Are we supposed to have Nvidia and AMD forever until AMD gives up because it becomes too expensive to compete with Nvidia? I’d like it to be someone else than Intel but as long as the price comes down I don’t care who brings it down.
And to be clear, if Intels new strategy is keeping the prices as they are I’m all for “fuck Intel”.
This is a big part of it, imo. They kissed the ring.
The other part of it is that, per the article, this is an “AI” pivot. This is not them making more consumer-oriented GPUs. Which is frustrating, because they absolutely could be a viable competitor in low-mid tier if they wanted to. But “AI” is (for now) much more lucrative. We’ll see how long that lasts.
CPU overhead is quite well known and actually damages a lot the arc cards’ position on the budget class
TropicalDingdongNot gonna make a lick of difference without the support to run CUDA.
woelkchenIntel GPU support?
ZLUDA previously supported Intel GPUs, but not currently. It is possible to revive the Intel backend. The development team is focusing on high‑quality AMD GPU support and welcomes contributions.
Anyways, no actual AI company is going to buy $100M of AI cards just to run all of their software through an unfinished community made translation layer, no matter how good it becomes.
OneAPI is decent, but apparently usually fairly cumbersome to work with and people prefer to write software in cuda as it’s the industry standard (and the standard in academia)
From what I’ve read about the “quality” of their drivers, .. NVidia isn’t under any threat, whatsoever.
Years before bugs get fixed, etc..
( Linux, not MS-Windows, but it’s Linux where the big compute gets done, so that’s relevant )
https://www.phoronix.com/review/llama-cpp-vulkan-eoy2025/5
for some relevant graphs: Intel isn’t a real competitor, & while they may work to change that .. that lag is SERIOUSLY bad, behind NVidia.
_ /\ _
Aren’t TPUs like dramatically better for any AI workload?
Intel’s Gaudi 3 datacenter GPU from late 2024 advertises about 1800 tops in fp8, at 3.1 tops/w. Google’s mid 2025 TPU v7 advertises 4600 tops fp8, at 4.7 tops/w. Which is a difference, but not that dramatic of one. The reason it is so small is that GPUs are basically TPUs already; almost as much die space as is allocated to actual shader units is allocated to matrix accelerators. I have heard anecdotally.
At scale the power efficiency is probably really important though
Yes, it works out to a ton of power and money, but on the other hand, 2x the computation could be like a few percent better in results. so it’s often a thing of orders of magnitude, because that’s what is needed for a sufficiently noticeable difference in use.
basing things on theoretical tops is also not particularly equivalent to performance in actual use, it just gives a very general idea of a perfect workload.
Good luck fucking things up like you always do
MentalEdgeWut?
Alchemist and Battlemage cards were fine.
Edit: oh no. It’s a pivot to AI compute 🤦♂️
DevolutionYou mean non shit non arcs? They tried already and failed already with battle mage.
talI don’t know if “GPUs” is the right term, but the only area where we’re seeing large gains in computational capacity now is in parallel compute, so I’d imagine that if Intel intends to be doing high performance computation stuff moving forward, they probably want to be doing parallel compute too.
If what‽
English clearly isn’t their first language, but the intent is pretty obviously “As if they aren’t already making ARC GPUs?”
clif
Tom Peterson, who’s been working at Intel for half a decade: Am I a joke to you
RememberTheApollo_Oh great, some wildly overpriced and underperforming GPUs.
They won’t be for you.
they’ve been quite good on the pricing front?
PrivateNoobAtleast it’s a 3rd contender, but afaik the Arc series had a decent enough pricing, although AMD’s prices seemed better, but not sure
ApplyingAutomation