Rumoured “abysmal 10% yield” for Intel 18A is fake news

Intel’s 10% yield rate rumours for its 18A lithography node are fake news – Here’s why

In recent days, there have been wave articles circulating online about Intel’s “abysmal 10% yield” for its Intel 18A lithography node. Today, we are here to say that these reports are misleading and that specific “yield” numbers are not what analysts should be looking at. For starters, reports on yield rates are meaningless without knowing the size of the chip being produced. This shows a fundamental misunderstanding that many reporters and analysts have when discussing this topic.

Defect Density matter, not yields

Chip size matters, and so does defect density. Silicon wafers are a specific size, and only so many chips can fit into a single wafer. The larger your chip is, the fewer chips you will be able to fit into a silicon wafer. The larger a lithography node’s defect density is, the more defects there will be on a silicon wafer.

When creating small chips, defect density matters less than on larger chips. This is why new lithography nodes are often used to create smaller chips first, like those used in smartphones. For smaller chips, defects will impact a smaller percentage of silicon dies than larger chips. This means that the impact of defect density is larger for bigger silicon dies. This is a primary reason why larger CPU designs are taking a “chiplet” design approach, as smaller chips have higher “yield” rates.

Let’s put things into simplified terms. If we know defects will ruin 20 small areas of a (hypothetical )silicon wafer during manufacturing, what would be the yield rate for small and large chip designs? If we have a small chip design where we can fit 100 into this wafer, only 20 will be harmed by defects, and the remaining 80 will be fine. That’s an 80% yield rate. If we build a larger chip and only 25 can fit into a wafer, we would have 20 defective dies and 5 good dies. That’s a yield rate of 20%.

To put it simply, a yield number without context is meaningless. Any yield number can be justified if you have a small enough or large enough chip design. If you want to see this effect in action, use a Die Yeild calculator like this one from isine.com.

Knowledgable analysts are calling this fake news

Patrick Moorhead, a well-known and respected tech analyst, has called the reporting on Intel’s 10% defect rate “fake news”. Below, former Intel CEO Pat Gelsinger thanked him for setting the record straight and said that he is proud of the work of the Intel 18A team.

thank you Pat for helping to set the record straight. I'm so very proud of the TD/18A team for the incredible work and progress they are making.

— Pat Gelsinger (@PGelsinger) December 7, 2024

Intel has been transparent when reporting on its 18A node’s defect rate

Back in September, Intel commented on the defect density of its 18A lithography node. Intel stated that “18A defect density already at D0 <0.40”, and that the “economics are right for us to make the transition [to 18A] now.”

Intel has confirmed that it has powered on and booted 18A-based chips and that they are “on track for launch in 2025.” Intel plans to use its 18A node to create its upcoming Panther Lake and Clearwater Forest series CPUs. That wouldn’t be happening if yields were an “abysmal 10%.”

Analysts have already used Intel’s reported defect rate numbers to calculate potential yields for chips. For example, the Tweet below states that they expect yield rates of over 60% on Panther Lake (PTL) chips. This assumes that Intel has not improved their 18A defect rate ahead of mass production.

Anyone with any sense did the calculations months ago, but it seems people misinterpreting yield vs defect rate will happen every single time.

— 𝐷𝑟. 𝐼𝑎𝑛 𝐶𝑢𝑡𝑟𝑒𝑠𝑠 (@IanCutress) December 6, 2024

Why are these “abysmal 10% yield” rumours for Intel 18A circulating?

There is a lot of financial interest in Intel right now, and there are many reasons why people would spread rumours about poor manufacturing yields. For starters, there is no doubt there are many investors shorting Intel stock. Beyond that, some want to see Intel fail for other reasons. There will also be Intel foundry competitors who would not like to see Intel Foundry Services succeed. This is just speculation, but there is no shortage of people who would give false information to news sources.

Hopefully, Intel will release a statement soon regarding its 18A lithography node. If not, rumours like this will continue to circulate.

You can join the discussion on Intel’s 18A “10% yield” rumours on the OC3D Forums.