Nvidia is fabless.

What Intel really needs to do is get competitive with TSMC. Gelsinger started the process instantly when he became the CEO, but it takes time.

It takes years to build a new fab. The market has no patience for that.

Intel is the cheapest tech company by a huge margin if look at book value.

Those fabs are expensive, but even if nobody wants Intel processors anymore, they could even manufacture Ryzen's.

The downside is that it takes 3-5 years to make profit on a fab, it's easier to be NVidia in that sense, but I wonder if it's sustainable in the long run.

no exponential growth is sustainable in the long run

They would need a culture change. Starting with a mostly independent Skunk Works division and a rethink of their software stack. I wouldn't bet on any of that happening.

Not convinced Intel can turn this around. Both on server and consumer side they've been losing share to AMD for like 7 years straight...cross generations, cross platform...just everywhere with amazing consistency. And if half of the AMD roadmaps/rumours are true then the next couple years will be worse.

...so yeah some casual R&D isn't gonna cut it...they need to pull a rabbit out of a hat urgently.

NVIDIA’s announcement included the declaration of a new Industrial Revolution based on AI. We’re coming to peak hype any month now. What will they ride when that wave(/bubble) dissipates?

(Microsoft announcing a few days ago that they have taken the lead in a new computing paradigm with their AI-infested devices was genuinely fascinating to me in that respect. The actual AI stuff that got added, I was like, (a) perhaps marginally useful for certain folks but (b) I don’t want any of it, and (c) it’s just literally the same stuff pushed for the past two years, only now with the same breathlessness as when people were genuinely wondering if ChatGPT was sentient (lol). Like, the huge pushback iTerm2’s creator got over adding an AI facility to its latest release puzzled some folks, but it really does feel like people are sick of the AI hype. AI evangelists are going to induce another AI winter if they don’t pump the brakes and start managing expectations better.)

Of course, AMD needed to pull a bigger rabbit out of a smaller and more decrepit hat before Ryzen.

Indeed. The fact that we all remember that as a remarkable feat shows just how dicey things are for intel.

It’s good that there is precedent proving it can be done, but duplicating it is another matter

Intels roadmap isnt so bad either. Its just about whether they can deliver

Yeah they’ll be sorta ok. I just wouldn’t put any money into what smells of decline to me