According to Reuters, Intel's Panther Lake PC chip is stuck in yield hell, with internal test data showing it's nowhere near commercial viability.

Despite all the talk about launching the chip at scale in 2025, two sources told Reuters that only a tiny fraction of Panther Lake silicon is usable. They called the whole project a “Hail Mary,” which doesn't exactly scream confidence.

Troubled Chipzilla has spent billions on 18A, upgrading fabs and building new ones in a desperate push to close the gap with TSMC. The plan was to crank out in-house chips and spruce up its foundry business. That pitch is now wobbling, as the company struggles to make Panther Lake profitable, let alone attractive to outside clients.

The chip includes fancy new transistors and a reworked power delivery system, which seem to be giving engineers grief. Intel’s usual threshold for ramping production is at least 50 per cent yield. Three sources told Reuters the company doesn’t hit solid profit margins until that figure hits 70 to 80 per cent.

Panther Lake’s yield is so low it would have to be sold at a steep discount or potentially even at a loss if they dare ship it, said the two sources with knowledge of the test data.

New chief executive Lip-Bu Tan has been poking at a strategic overhaul of Intel’s struggling foundry business, hoping that success with 18A might spark external interest. But with the chip yields in tatters, the company's entire advanced manufacturing effort is under pressure.

Chipzilla has warned it could pull out of cutting-edge manufacturing if it can't land clients for 14A, the even more ambitious successor to 18A. Without customers or working silicon, that foundry future starts to look like science fiction.