Significant progress in Intel 14A: Defect reduction rate reaches best level since 22nm.

Intel's 14A (1.4nm) process has achieved a faster-than-expected decrease in defect density, marking its best performance since 22nm. CFO David Zinsner stated at Deutsche Bank's 2026 Technology Conference, "We haven't seen this kind of performance since 22nm."

The 14A is planned to enter risk production in the second half of 2027 and mass production in 2028. With about two years to go before mass production, there have been substantial changes on the customer side: Intel's internal design team has begun developing products on the 14A, and external foundry customers have shifted from evaluating technical data to inquiring about capacity allocation.

Comparing to 22nm: Good News and Triple Limitations

"The defect density of 14A is better than our target curve, and the defect rate of decline is also better than any previous node. We haven't seen this performance since 22nm—which is arguably one of the best processes Intel has ever launched," Zinsner said at the conference.

22nm was Intel's first FinFET process, leading the industry at its launch—competitors didn't follow until the 14/16nm nodes of 2014-2015. The Ivy Bridge processor was a huge success after its launch in 2012, and 22nm served Intel's CPU production line until 2016, after which it continued to be used in other product lines.

However, Zinsner's comparison has strict boundaries. He benchmarked the defect reduction trajectory of 14A about two years before mass production against the performance of 22nm around the same time in 2010, rather than a direct comparison of the absolute defect levels of the two generations of processes. At the same time, the advancements in wafer inspection equipment over the past 16 years have changed the measurement of "defects," and defect density itself is not directly equivalent to yield.

The complexity of the 14A process makes these data even more compelling: second-generation gate-around-the-loop (GAA) RibbonFET transistors, second-generation PowerDirect power supply, and High-NA EUV lithography, making the technology far more challenging than 22nm. The ability to rapidly reduce defect density with such a process combination is particularly noteworthy in Intel's recent manufacturing record—previously, 14nm was delayed by a year due to insufficient yield, the initial 10nm process failed, 20A was cancelled, 18A still had high defect rates during mass production, and Intel 4 and Intel 3 saw almost no publicly disclosed progress.

Clients: From "Looking at Data" to "Demanding Production Capacity"

Zinsner revealed that Intel's internal design team has already developed products on the 14A. He described these internal customers as "probably the most discerning of all," adding, "Their decision to design products on the 14A is a huge boost to our confidence."

The attitudes of external contract manufacturers are also undergoing a qualitative change. According to Zinsner, CEO Chen Liwu and his team now meet with clients weekly. "Clients have shifted from looking at data to asking, 'How much capacity can I get? How will the supply schedule be arranged?' We now have confidence in our external clients for 14A."

The 14A is still about two years away from mass production. Intel says the node is expected to become a long-lived process, like the 22nm process.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.