It was a strategic mistake for Intel to even get into dGPU space. GPU is arguably the #3 priority at Intel behind CPU and foundry. Where as it's #1 priority at NVIDIA and #2 at AMD behind CPU. What makes Intel think it could catch up to existing leaders by investing less? It might be feasible if Intel's goal was to become a cheap #3 option in dGPU with all the dGPU produced by Intel's own fabs. Then it could help Intel convince NVIDIA or AMD that it's capable of fabbing GPU's for them. Intel could even spin off the dGPU design team while continue to have them as a foundry customer. Yet Arc Alchemist uses TSMC, so it really does nothing for Intel.
8 replies (most recent on top)
AXG was not a mistake it's actually a great revenue stream if Intel could of got it right. The issue with AXG was the leadership, same individuals making the same mistakes since Haswell. Always over budget, never meeting commits and always have an excuse as to why they can do better. Raja, Lisa and Jeff all needed to be replaced and it may of stood a chance
“ posted it here without my permission”
lol, Better Call Saul to get some justice from someone posting your IP.
OP, you copied my comments from Blind and posted it here without my permission. I don't appreciate it. My comment was a response to a specific thread. I have no interest in starting a new thread in this forum to criticize Intel.
Graphics was clearly a mistake from the beginning. I guess they might have taken some slice of the pie, but they were never going to 'win'.
That's Intel's future, not to win segments but to just take a sliver enough to justify its existence
If you thing Intel winning in graphics is hard then wait until you really understand what it means to be a foundry for other companies silicon... Here is a brief list and Intel is behind on all of of it:
- Low cost manufacturing
- Process leadership
- Broad ecosystem of tools and up
- Support for multiple legacy nodes for customer product longevity
- Ability to manage high sku count and efficiently shuttle/load/schedule/crossover fabs
- Ability to convince customers you won't compete with them
- Willingness to serve customers instead of insulting them.
As you can see, this set of skills is something Intel doesn't have. Graphics market entry should have been possibly for Intel but IDM 2.0 is not going to work.
The barbarians have already over run the gates and are ra-ing and pilliagering.
AMD will surely cross over 50% ether with their current momentum and superior management, talent, execution, and , technology.
Nvidia already gave Intel a can of whoop a$$ and will continue for foreseeable future given their far superior management, vision, talent, execution and technology.
Intel is worried about the data center.
Nvidia has a path to bundle cpu+gpu and cut out Intel with a much bigger R&D budget and a superior PPA with TSMC.
They almost bought ARM to drive this strategy.
It’s not just AMD that wants to eat our lunch.
The barbarians are at the gate folks and they’re hungry.
Let’s be serious where has Intel been successful entering the market as a distant third with little history in the business and offering cheap alternative.
Intel has little strategic or business reason to get into graphics. Many years ago the argument was that better graphics enabled more rich experience required more powerful x86 CPU and would sell more x86 thus could be money losing.
Currently one could argue with leading edge silicon it could leverage IDM inherent advantage and put more volume into there fabs. Oops two strikes Intel silicon technology is two generations behind, it has little fabs compared to who the competitors use. The last joke is the GPUs aren’t even made internally as they would be so poor performing and money losing.
Intel Graphics currently is a distraction and sucks valuable design and management resources always from their broken core business in x86 and IDM manufacturing. Every review of graphics is valuable time not focused on the bleeding in x86, silicon technology and manufacture, totally FUBAR strategy.