Do you think Intel can find its way back to the top or are those days behind us for good?
7 replies (most recent on top)
2020 is the year the wheels fall off completely.
TSMC is ramping to HVM for 5nm in Q2. It's pretty sad that TSMC's 5nm defect density volume may surpass Intel's struggling 10nm node. Customers have already taped-out on TSMC 5nm.
https://www.anandtech.com/show/15016/tsmc-5nm-on-track-for-q2-2020-hvm-will-ramp-faster-than-7nm
https://wccftech.com/tsmc-producing-5nm-asics-for-both-bitmain-and-canaan-in-2020
Try asking again in 3 years
It is hardly possible to change culture after so many years. Intel will have to spin off fabs at the end. It is not possible to be competitive when you have only one in-house customer.
10 years ago it was not so important but today fab market is very competitive and therefore Intel fabs lack behind.
It's a shame to see that today Intel is struggling with 10nm while competition switched to 7nm long ago.
Intel's business model have structural issues.
It's over unless AMD is stupid and shoots themselves in the foot with another c-appy architectural design (ie. Bulldozer). Intel's advantage over the decades has been the 3 to 5 year lead on manufacturing tech. The manufacturing ultimately influences the design via transistor budget. The key point here is that Intel's manufacturing tech was developing at the same rate as the rest of the industry. The lead didn't increase from 3 years to 10 years. It stayed constant. The advantage came from being a first mover. Now that Intel has fallen behind in manufacturing it will be almost impossible to jump ahead. The best hope now is to just catch up with TSMC.
LOL. That's a good one OP. Save it for 4/1
Maybe 2021. 2020 will be a year of losing mindshare, marketshare, and margins. Only place I see little to no impact will be the mobile/ laptop market. AMD has never made a competitive offering superior to intel there. AMD has a superior transistor and fab partner so that may also change 2020, chiplet design comes with a minor power penalty(100mW) but you would think with better throughput per Watt AMD could offset that.
My rant/
Intel's biggest hurdle is its antiquated fabs and fab systems. It is a dinosaur that is neither nimble nor flexible with a work force that has no understanding of automation and solution is to brute force manpower to solve all problems even when apparent that it is a waste of resources. Intel's inefficient fab culture is another reason for their decline and slowing down their ability to compete, losing there own benchmarking to external fabs and shifting designs and fabrication out of house. Losing vertical integration because an external fab executes better and does it for cheaper. Intel may be able to design their way out from being behind it's competitors (ARM, AMD, Nvidia) but the fabs will need to be reborn to survive external benchmarking.