FDC Pat Gelsinger wants to make Intel the world's second largest chip manufacturer by 2030, and that means serving businesses the x86 giant has traditionally seen as competitors.

"We want to be the foundry for the world. If we're going to be the Western foundry at scale, we can't be discriminating in who's participating in that," he declared during Intel's Foundry Direct Connect event in San Jose, California Wednesday.

To do this, Gelsinger is splitting the chip shop down the middle, to form two independent organizations under the Intel brand. As part of the reorganization and rebrand, Intel Foundry Services will be expanded to include the chipmaker's technology development, supply chains, fabrication, and packaging services under the IFS moniker.

Meanwhile, Intel's Product division will focus its efforts on development and licensing client, desktop, and networking kit. In effect, this Intel products arm will function more like a fabless chipmaker than it has previously.

These organizations will be legally distinct and have their own staff and processes with minimal overlap to ensure confidentiality, explained IFS head Stu Pann.

"We're very disciplined about this. We have two separate sales forces. We're building two separate ERP systems. We're outlining two separate legal entities," he said. "Intel Foundry will do arm's length transactions with the Product groups."

The Product group will also get fab allocations - which he referred to as capacity corridors - the same as any other customer. He did admit that Intel Products would likely drive the bulk of Intel's business for the next couple of years.

Perhaps more importantly, Intel Foundry will maintain strict confidentiality with each of its customers - regardless of whether they're internal or external. "This is at the heart of what it takes to be successful in foundries," Intel EVP of manufacturing Keyvan Esfarjani explained.

'Strange bedfellows'

While Intel has legal frameworks in place to separate the Foundry and Product divisions, the strategy is already playing out with Intel's partnership with Arm.

Arm is working to make its intellectual property available on Chipzilla's process nodes. For example, Faraday is already announcing chips using both of their technologies.

The unorthodox nature of the partnership wasn't lost on Arm CEO Rene Haas, who, joining Pann on stage, described the coupling as "strange bedfellows."

"I was trying to think of a parallel relevant to this story and the only thing I can think of is, for those who can harken back, is when Walt Mossberg asked Steve Jobs what it was like to see iTunes running on Windows? I think he said it was like ice water in hell, but I won't go that far," Haas said.

The foundry team's objective is to fill the fabs. We hope that includes Jensen and Cristiano, and Sundar, and you heard today it includes Satya, and I even hope that includes Lisa going forward,

Alongside Arm, Gelsinger talked up its strategic partnerships with chip design companies like Cadence, Ansys, Siemens, and Synopsys, and touted a contract from Microsoft to build chips on its 18A process node. Microsoft - a relative newcomer to custom silicon - is only the tip of the iceberg for Gelsigner's ambition.

"The foundry team's objective is to fill the fabs. We hope that includes Jensen and Cristiano, and Sundar, and you heard today it includes Satya, and I even hope that includes Lisa going forward," he said, listing off executives from Nvidia, Qualcomm, Google, Microsoft, and AMD.

But while Intel is attempting to preempt any conflicts of interest through this reorganization, Gelsinger doesn't appear to be ready to give up the keys to the kingdom and spin off its foundry business as a standalone entity.

That said, the spinoff of Intel's programmable solutions group last year suggests that the same could happen once Intel Foundry is more established.

Intel plots course for new process tech

Of course, attracting major chip vendors like Nvidia or AMD first requires having a process and packaging tech worth the risk.

To this end, Intel teased its next-gen process tech which, from the roadmap, should start rolling out between 2025 and 2027. Among the nodes is a successor to the unreleased 18A process node called Intel 14A, which will be its first to use ASML's High-NA extreme-ultraviolet lithography tech.

There are actually two process nodes based on the tech: 14A and an enhanced version called 14A-E, which will presumably come later. In addition to the process shrink, Intel also plans to roll out a performance-optimized version of its 18A node, predictably named 18A-P as well as enhanced versions of its older process nodes including Intel 3-E and Intel 3-PT. 

The latter is a performance optimized process node designed with through-silicon VIAs for applications employing 3D packaging. Intel plans to put this tech to work on its next-gen many-cored CPU code named Clearwater Forest.

We don't know much about the chip just yet, but we do know it will stack CPU tiles built on Intel 18A on top of a carrier tile based on Intel 3. We've seen a similar technique employed on AMD's MI300A APUs, though the packaging tech is a little different.

Intel believes this process tech, along with globally distributed and sustainable fabs, will form the basis of what Gelsinger called a "systems foundry."

The idea seems to be that Intel will provide more than just manufacturing and packaging services like TSMC. Rather it will promise to provide other value adds to support more complex heterogeneous chiplet designs using internally and externally sourced intellectual property.

Still waiting

Since Gelsinger first announced Intel would open its foundries to contract manufacturing in 2021, the chipmaker has committed to building more than $100 billion worth of chip factories across the US and Europe.

Much of this work, however, hinges on substantial government subsidies under the US and EU CHIPS Bills. But despite Global Foundry's $1.5 billion award Monday and commerce secretary Gina Raimondo's virtual appearance during Wednesday's keynote, Intel is still waiting for its slice of the pie.

The wait could yet be worth it. As we recently learned, Intel could receive as much as $10 billion in subsidies from the $39 billion set aside for fab projects. And as we've previously reported, that number could end up being even higher.

On the bright side - for Intel at least - there could be even more funding on the horizon. When asked about the prospect of a second CHIPS bill, Raimondo didn't nix the idea. "I suspect there will have to be … CHIPS two or something else … continued investment … if we want to lead the world," she replied.

As tantalizing a prospect as CHIPS cash may be, the speed or lack thereof with which the first bill has been dispersed has already caused issues for Intel's fab build out. Earlier this month, it was reported that Intel had delayed its Ohio foundry project to 2026 due to the weak semiconductor market and delays in receiving CHIPS Act subsidy cash. ®

https://www.theregister.com//2024/02/22/gelsinger_intel_split/