We’ve already covered the first big announcements from the Intel Foundry Direct Connect event, but it turns out that Intel misspoke about a presentation that was thought to be under NDA—now Intel says the presentation is open to the public for reporting, so we have some more news to share: Intel’s previously unannounced Intel 10A (1nm compliant) in production/development in late 2027 will enter, marking the arrival of the company’s first 1nm node, and its 14A (1.4nm) node entering production. 2026. The company is also working on building fully autonomous AI-powered fabs in the future.
Intel’s Keyvan Esfarjani, the company’s EVP and GM and foundry manufacturing and supply, conducted a very insightful session covering the company’s latest developments and how the roadmap unfolds in the coming years. Is.
Here, we can see two charts plotting the company’s K-WSPW (thousands of wafer starts per week) for Intel’s various process nodes. You’ll notice that there is no label for the Y-axis, which would give us a direct read on Intel’s production volumes. However, it does give us a solid idea of Intel’s planned node production rate over the next several years.
Intel hasn’t specified an arrival date for its upcoming 14A node in its previous announcements, but here, the company indicates that it will begin production of the Intel 14A node in 2026.
More importantly, Intel will begin production/development of its as-yet-unannounced 10A node in late 2027, filling out its roster of nodes built with EUV technology. Intel’s ‘A’ suffix represents Angstroms in its node naming convention, and 10 Angstroms converts to 1nm, meaning this is the company’s first 1nm-class node.
Intel hasn’t shared any details about the 10A/1nm node but has told us that it classifies the new node as having at least a double-digit power/performance improvement. Intel CEO Pete Gelsinger has told us that the cutoff for the new node is about a 14% to 15% improvement, so we can expect the 10A to have at least that level of improvement over the 14A node. (For example, the difference between Intel 7 and Intel 4 was a 15% improvement.)
The capacity for Intel 4 and Intel 3 processes isn’t quite as fast as 20A/18A, but that’s not surprising – the majority of wins for the company’s third-party foundry business have been with the 18A node, which Intel says Is. is according to plan. As shown, Intel will also steadily reduce its overall production of its 14nm, 10nm, Intel 7, and 12nm nodes as it transitions to EUV-enabled nodes.
Notably, Intel’s slide has a footnote that says, “*Roadmap: Final scale, pace and implementation depend on business conditions and incentives,” mirroring the company’s ongoing statements that funding from the CHIPS Act It will affect the ability to scale production.
Intel’s 18A and 20A nodes have been in some form of production since at least 2023, which isn’t too surprising — the company won a RAMP-C contract with the US government for the 18A in 2021, and IBM, Microsoft, And companies like Nvidia have been working on test chips ever since (understandably Intel won’t share details of the program with the public, but it won a $1 billion award from the US government for the program).
Additionally, Intel’s 20A allows the company to simultaneously integrate two new technologies – Backside Power (PowerVIA) and GAA Transistors (RibbonFET). To de-risk the process and avoid stumbles like it saw with 10nm, Intel announced in June 2023 that it ran two different flavors of 18A through its fabs, one with GAA only, while The other had backside power only, to ensure both. Features used to work separately before merging into a single node. As such, we can expect 20A wafers of various types to be flowing through Intel fabs for quite some time.
As shown at right, Intel will also aggressively expand its advanced packaging production capacity for Foveros, EMIB, SIP (Silicon Photonics), and HBI (Hybrid Bond Interconnect). Advanced packaging capability has been a key choke point for AI accelerators’ current shortage. This increased capacity will ensure a continuous supply of advanced processors with complex packaging, including HBM.
Intel’s ramp to high-capacity packaging is explosive — the company had little production capacity for these interconnects in 2023. It is now on state-of-the-art packaging and will use OSATs (Outsourced Assembly and Test Companies) for standard packaging tasks.
The second slide in the album above visualizes how Intel’s move to act as an external foundry will allow it to increase both the amount of production for each of its nodes and the time each node spends in production. will, thereby maximizing profit from its fab. and equipment costs as it serves its customer orders over a longer period of time.
Asfarjani also shared details about Intel’s global operations. In addition to its existing facilities, the company plans to invest $100 billion in expansion and new production sites over the next five years.
The slides above outline the various Node production locations, with 18A taking place at Fab 52 and 62 in Arizona. In contrast, the advanced packaging and 65nm foundry operations for the tower will be at Fab 9 and 11X in New Mexico. Intel did not share where it plans to manufacture its 10A node, and it also has ongoing expansion in Ohio, Israel, Germany, Malaysia and Poland.
This geographically distributed manufacturing capability, spanning both chipmaking and packaging, allows Intel global redundancy in its operations while allowing its foundry customers to benefit from a fully US-based supply chain. Also offers options.
As we illustrated in our coverage of our tour of Intel’s Penang, Malaysia facility, the company relies heavily on automation in its foundries. Intel now plans to use AI in all parts of its production flow, from capacity planning and forecasting to optimization and actual floor-level production operations, in a ’10X moonshot’ effort.
Asfarjani did not provide a timeline for the company’s moonshot effort but said it will affect every aspect of its operations in the future. This includes the introduction of AI “Cobots,” which are collaborative robots that can work alongside humans, and widespread robotic automation in manufacturing processes.
In the meantime, Intel will continue to aggressively pursue any and all potential customers for its operations. You can read more about these efforts in our interview with Stu Pann, SVP and GM of Intel Foundry Services, who is tasked with making Intel Foundry the second largest foundry in the world by 2030.