Intel's Foundry division is near to sealing a deal for its advanced packaging technology that would contribute billions of dollars a year to the struggling chipmaker, CFO David Zinsner said on Wednesday.
"We've gotten really good engagement from customers on this business," he said during the Morgan Stanley Technology, Media and Telecom conference on Wednesday. "We're close to closing some deals that are in the billions of dollars per year in terms of revenue."
Advanced packaging has become a focal point of semiconductor manufacturing amid the AI boom as accelerators have grown more complex, meaning many GPUs are composed of multiple compute and memory dies that must be fused together.
Intel has invested heavily in its EMIB and Foveros 2.5D and 3D packaging tech to support multi-die processors like its Xeon CPUs and (now discontinued) Ponte Vecchio family of GPU Max accelerators. Perhaps more importantly, that tech isn't limited to chips Intel makes in its own fabs – the company has already used it to meld silicon manufactured in-house with chips made by TSMC.
This is likely welcome news for shareholders, who've watched in dismay as Intel's Foundry Division has cost the company billions each quarter. Of course, there's no guarantee those deals will materialize, and even if they do, Intel will need to deliver on time. The company doesn't have the greatest reputation on that front, we'll note.
Alongside advanced packaging demand, Intel CEO Lip-Bu Tan appears to have changed his tune regarding the external adoption of its 18A process tech.
The chief executive, who celebrates his first year in the role later this month, had previously considered 18A for Intel’s own use only. Intel’s foundry biz planned to offer its next-gen manufacturing process, 14A, as its first mainstream commercial offering.
But according to Zinsner, Tan's opinion of 18A changed over the past year.
"While Lip-Bu was thinking that we probably should focus on 14A as a foundry node and make 18A just an internal node, now that we've seen some real progress there, I think he's now starting to recognize that this is actually a good node to offer to external customers as well," he said.
The first products based on Intel's 18A process tech, codenamed Panther Lake, began shipping in PCs and notebooks earlier this year, and it seems to have caught the attention of potential customers.
"We've been getting some inbound interest in 18A-P as a foundry node," Zinsner said, but didn’t name potential customers.
Intel 18A-P is an upgraded version of 18A that the company has suggested will deliver eight percent higher performance per watt, which compares favorably with TSMC's competing 2nm process.
Despite offering good news for Foundry’s fortunes, Zinsner said the biz is still at least a year away from breaking even.
"I think we still feel good about our original guidance, which is we expect to exit '27 at break-even operating margins for Foundry," he reiterated at Morgan Stanley’s event.
The CFO said one thing that could derail that forecast would be higher-than-expected external demand, as that would mean Intel needs to spend more to build additional manufacturing capacity.
“It'll be a good problem to have," Zinsner said.
®
Source: The register