Intel’s ambitious 18A chip manufacturing process-once hailed as its comeback weapon against TSMC-is reportedly on the chopping block for external customers, according to a new Reuters report. 
The technology, positioned as Intel’s answer to TSMC’s cutting-edge 2nm node, now risks becoming an internal-only tool, raising serious doubts about the company’s foundry strategy.
The 18A node (1.8nm-class) was supposed to showcase Intel’s future-facing RibbonFET and PowerVia technologies, and former CEO Pat Gelsinger even staked the company’s future on it. But half of 2025 has passed with little progress revealed, and insiders suggest new CEO Lip-Bu Tan is losing faith in its market appeal. If the rumors are accurate, Intel will still use 18A for in-house chips, such as those for Amazon and Microsoft, but external customers may be left behind-pushing the dream of Intel as a TSMC competitor even further out of reach.
TSMC’s 2nm process is ramping up for mass production this year, with Apple as the likely first client. Meanwhile, Intel’s 18A has lagged in key metrics like transistor density, and Tan reportedly believes it’s simply not worth the cost or the risk anymore. Scrapping it for outsiders could mean millions in write-offs, but perhaps less damage than an underperforming product would cause.
In its official response, Intel didn’t deny the report but offered a polished corporate spin: “Lip-Bu and the executive team are committed to strengthening our roadmap, building trust with our customers, and improving our financial position for the future.”
So what’s next? Insiders hint that Intel might pivot to the 14A node instead, though that would be a step back in prestige, if not practicality. As for a foundry spin-off? No confirmation yet. But if 18A becomes internal-only, Intel’s hopes of luring big foundry clients like Qualcomm or Nvidia will face a serious credibility gap.
While some still believe Intel can recover, for many observers, this looks like another chapter in the company’s painful struggle to reclaim relevance in the foundry wars-one failed node at a time.