LTD, 18A failings, future

Ironical that Kevin B and others are gloating how great 18A and Clearwater Forest are, when in reality:

18A yield is in extreme poor health and there is absolutely no denying of it. The chance of catching up to TSMC’s 2nm next year is 0%

There is no interest by external foundry customers to ink any deals. The PDK 1.0 does NOT meet the needs; no believable timeline exists for PDK 1.1

18A is very expensive compared to N3E. When N3E has better performance over 18A, it makes zero sense to sign up for 18A which is poor yielding and significantly more expensive

Given TSMC’s N2 to be ready by early 2026, it is difficult to make a case for 18A now. Perhaps there might be something in there by end of 2026, but then N3E will have made more improvements, and N2 would be 10-15% better than N3E.

For the above reasons, 18A maybe regarded as just a process Dead on Arrival. No two ways around it.

Who should take the blame? Two individuals: Ann and and Ryan. Expect mass exodus from LTD in the coming weeks. Let’s all wish these 2 individuals the best as they try to keep feeding koolaid to stay afloat.

Comment! It's anonymous!

11 replies (most recent on top)

Op is right- Known by many that 18A yields are NG. It fries me to hear Pat lie, (just like BK did), telling the world it's 'ontrack'.
Just wait and see... More shall be revealed!

The problem for Intel is that it's almost impossible to compete on cost and quality compared to labor from Asia. Even TSMC can't make their new Arizona fabs competitive, against their own Taiwan fabs, using US labor. Even with the US government applying pressure and incentives for them to build a fab here they really don't want to do it and are dragging their feet. It's really no different than any other manufacturing in the US compared to Asia. Look at Tesla Fremont vs Tesla Shanghai. It's a night and day difference just using a different workforce.

Couldn't you argue that the semiconductor industry scaling is close to the end, regardless of company? Someone has 3nm, and someone else 7nm, and working on 1.8nm .... is there really many other places to go with transistor size scaling. Maybe it's time for big gains using other new design and technology ideas?

Apple unveils M3, M3 Pro, and M3 Max, the most advanced chips for a personal computer
The industry’s first 3-nanometer chips for a personal computer debut a next-generation GPU architecture and deliver dramatic performance improvements, a faster CPU and Neural Engine, and support for more unified memory. How's does 18A stack up against this.

Apples move to their own desktop chips looks great now. M3 all the way.

The bottom line is if 18A does not have better price/performance than TSMC no reasonable customer will buy it. Intel cannot force a product down a customers throat anymore. Too many option are out there anymore. Before it was take my X86 at this price. Those days are over.

If this is all true, someone should be held accountable.

You got a source for your first point?

The boardrooms of Dell, Lenovo and HP are probably panicking right now and begging Lisa Su to supply them.

He-l, even Qualcomm won lots of seats at the table.

Expect Mediatek and Nvidia to join next year as well.

TSMC HVM for N2 will be much earlier than early 2026.

The correct restructure would be to layoff all G10 and up including ELT

Thread regarding Intel Corp. layoffs

LTD, 18A failings, future

11 replies (most recent on top)

Post a reply