Key Takeaways From Intel Foundry Direct Connect 2025

Forbes

30-04-2025

Intel CEO Lip-Bu Tan On Stage At Direct Connect.
Day one of Intel Foundry's Direct Connect event, which is currently underway in San Jose, California, just wrapped and I wanted to share some thoughts. The day was filled with interesting developments from both Intel itself and its diverse array of ecosystem partners. There was a time when no one in the semiconductor industry could build transistors like Intel, but a series of missteps and changes in strategy spread out over the better part of the last decade shifted that paradigm. Now, Intel is attempting to propel its manufacturing processes back into a leadership position, while simultaneously trying to woo fabless semiconductor customers to manufacture their chips in Intel's fabs. To do that, Intel has to establish more trust in the industry and not only prove the competitiveness of its processes, but also that partners can make the chips with familiar and established workflows, consistently and with high reliability. Just about everything I heard today made me believe Intel is on track to do just that.
Intel's Dr. Naga Chandrasekaran Holding A Intel 14A Test Wafer.
The day began with a keynote address by newly-minted CEO Lip-Bu Tan. Lip-Bu was refreshingly direct and to the point. He outlined a number of Intel Foundry's challenges, and what moves the company was making to address them and chart a successful course for Intel Foundry's future. What I found most assuring during Lip-Bu's time speaking, however, was the parade of partners he brought out on stage and what they had to say.
In today's advance semiconductor market, there is no single entity that can be successful on its own – there are simply too many technologies and packaging innovations at play in modern multi-chiplet devices. To that end, Lip-Bu brought out the CEOs of essentially all of the leading EDA tools providers.
Sassine Ghazi, Synopsys CEO, At Intel Foundry Direct Connect.
Sassine Ghazi of Synopsys, Cadence's Anirudh Devgan, Mike Ellow of Siemens EDA, and PDF/Solutions' John Kibarian came out in succession and all explained how their tools, IP and/or flows have been optimized for Intel's upcoming 18A process and the work is underway for future nodes, like 18A-P, 14A and its derivatives (among others). This strong commitment from Intel's ecosystem partners should carry a lot of weight; the overwhelming majority of fabless semiconductor players will be familiar with these tools, which should facilitate transitions to Intel Foundry's manufacturing processes.
All of the EDA providers also voiced their impressions of Intel 18A, which were universally positive. In fact, the scuttlebutt amongst industry insiders has been mostly positive regarding 18A for quite a while. Multiple conversations I've had with trusted folks from the automotive, mobile, and processor industries have all suggested Intel 18A is a highly competitive process. That said, there weren't any new customer announcements today. I suspect there will be more coming soon, however, once Intel Foundry begins talking more about Intel 18A-P, which is a higher performance derivative of 18A that's better suited to a wider array of applications. 18A is already a competitive process and Intel will be shipping its next-generation Panther Lake-based processors built on 18A in volume later this year, but 18A-P addresses some technical issues that should make it a better fit for a more diverse array of architectures.
Cadence President & CEO Anirudh Devgan At Direct Connect.
Anyone scrutinizing and paying close attention to Intel and its moves with Intel Foundry needs to understand some fundamental realities related to 18A. In addition to being a leading-edge process, with smaller features, it introduces two innovations – Ribbon FETs and PowerVia backside power delivery. It's hard enough to bring a new manufacturing process to market, let alone one with two new key technologies. And to realize the full capabilities of 18A requires chip designers to make specific considerations from the start. To move an existing design or a design that was already in-flight that was originally slated for an existing process to18A is possible, but it may not benefit from Ribbon FETs and to utilize backside power delivery, can incur additional cost. However, optimizing a design for 18A from the start will offset most of the cost related to backside power delivery by easing density requirements and eliminating the need for some metal layers on the frontside. It will also enable designers to realize the maximum benefits of Ribbon FETs, and optimize power, performance, and area.
Of course, chip designs don't happen overnight, so it's going to take some time for Intel Foundry's future customers to navigate these waters and decide if 18A, one of its derivatives, or a future process will be the best fit.
Intel 18A Details.
As for the processes themselves, Intel made plenty of encouraging claims. 18A has already achieved 95%+ of its targets and the optimization areas have already been identified to tweak the process to hit or exceed 100% of those targets next quarter. Intel 18A should deliver a >15% improvement in performance per watt, with 1.3x improvement in chip density versus Intel 3, with an additional ~8% improvement in perf-per-watt coming with 18A-P, along with support for 3D die stacking.
Intel 14A Disclosures.
Intel Foundry also disclosed some numbers related to its future 14A process. 14A is looking particularly strong at this early stage; Intel is expecting a 15-20% performance per watt increase versus 18A, with another 1.3x increase in density, and a 25-35% reduction in power. Intel 14A will also make use of High NA EUV and introduce second generation Ribbon FETs and PowerDirect. PowerDirect is an evolution of PowerVia that enables direct contact power delivery.
Intel Foundry Process Roadmap
There was a lot more disclosed during Intel Foundry Direct Connect, including a number of advancements in packing technology that improve interconnect density and performance, but it's just too much to cram into this report. Suffice it to say, after digesting day one's disclosures, my confidence in Intel's foundry efforts has reached a new high. The innovations, technologies, and tools to enable them appear to be in place. Now it's a matter of bringing in new customers, executing, and getting new products to market. Considering the company's huge investments, and the geopolitical pressures to bring more advanced semiconductor manufacturing back to the U.S., I think Intel Foundry can pull this off.