9 Takeaways from CEO Pat Gelsinger’s Intel IDM 2.0 Strategy Briefing
Channel Futures breaks down Intel’s “Engineering the Future” growth plan.
March 29, 2021
![Takeaway Takeaway](https://eu-images.contentstack.com/v3/assets/blt10e444bce2d36aa8/bltfa73a3a5f2387ed2/6524381896824aa6b7cfc109/Takeaway.jpg?width=700&auto=webp&quality=80&disable=upscale)
Shutterstock
When Intel initially designed its 7 nm processes, the company determined that the extreme ultraviolet-lithography (EUV) manufacturing technique was nascent. Because of that, Intel decided to the use of EUV. Consequently, that decision added more complexity to its manufacturing process, Gelsinger said. Intel has now “fully embraced” EUV, paving its way toward 7 nm processes.
“We’ve re architected and simplified our 7 nm process flow, increasing our use of EUV by more than 100%,” Gelsinger said. “We have a very strong partnership with ASML [which provides the EUV machinery to mass-produce integrated circuits], and our plans to now stay on the leading edge EUV usage are well underway.”
The current plan is to tape in Intel’s 7 nm compute tile for Meteor Lake in the second quarter of this year.
“Meteor Lake features a breakthrough new x86 architecture and modular design, utilizing multiple manufacturing processes across our XPU IPs,” he said.
Gelsinger showed a prototype of Intel’s first XPU, a discrete GPU for data center servers -based on Intel’s X-LP microarchitecture now used for Android-based cloud gaming and media streaming.
The XPU can push a peta-flop of compute onto a small piece of silicon. Code-named Ponte Vecchio, the high-performance silicon uses more than 40 different tiles integrated into a single package. The package combines more than 100 billion transistors manufactured using multiple process technologies.
Two key new packaging technologies and Intel, Ferveros and IDMB, will enable the company to bring the exascale GPUs to market:
· Ferveros is an advanced 3-D packaging technique that allows Intel to stack various tiles vertically, including high-performance CPU, graphics and AI processors for the first time. “Ferveros gives us the flexibility to mix and match various IPs, while optimizing for performance and power efficiency,” Gelsinger said.
· Embedded Multidye Interconnect Bridge (EMIB): Gelsinger described EMIB as “an elegant performant and cost-efficient way of connecting multiple heterogeneous tiles on a single package.”
Intel’s XPU vision includes its oneAPI toolkit, to provide unified interfaces for applications that run across multiple architectures including CPUs, GPUs and FPGAs.
The launch of Intel Foundry Services means the company is not only going to manufacture its own chips, but it will also produce them for other semiconductor companies, including its competitors.
While Intel could have made this move at any time, why is the company doing it now? Gelsinger said Intel conservatively sizes the foundry opportunity as a $100 billion addressable market by 2025.
Dr. Randhir Thakur will lead Intel Foundry Services, reporting to Gelsinger.
“This business unit will be completely dedicated to the success of its customers with full P&L responsibilities,” Gelsinger said. “This model will ensure that our foundry customers products will receive our utmost focus in terms of service technology enablement and capacity commitments.”
Intel Foundry Services can handle manufacturing needs for x86 cores, graphics, media display, AI, interconnect fabric, Arm and RISC-5. Gelsinger said partners including Amazon, Cisco, Ericsson, Google, IBM, Mac, Microsoft, Qualcomm, and others have endorsed the effort.
Besides selling capacity to others, Intel is expanding its use of third-party foundry capacity across its portfolio. Gelsinger said Intel will increase usage of foundries operated by Global Foundries, Samsung and TSMC.
“As we grow the business, we expect our engagement with foundries to grow that both size and scope,” Gelsinger said. “This includes manufacturing a range of modular tiles on advanced process technologies, including products at the core of our compute offerings for both client and data center segments. This will provide us with increased flexibility and scale we need to optimize our road maps for cost, performance, schedule and supply, giving us a unique competitive advantage.”
Intel is expanding its manufacturing capacity in the U.S. and Europe to provide less dependence on any specific region. Noting that 80% of leading-edge foundry capacity is concentrated in Asia, Gelsinger believes “the industry needs more geographically balanced manufacturing capacity.”
The U.S. expansion includes its $20 billion investments to build two new fabs at its Ocotillo, Arizona, campus. The fabs will provide added manufacturing for Intel’s own products as well as for Intel Foundry Services customers. Intel will also build another fab somewhere else in the U.S. in a location to be announced.
The company said its manufacturing expansion in the U.S. and Europe will:
· Result in the addition of 25,000 jobs in the U.S.
· Position Intel to compete for a Department of Defense commercial foundry designed to meet security requirements of the U.S. government.
· Shift the balance of its global manufacturing capacity.
The Intel-IBM partnership aims to tap both companies’ capabilities and expertise to collaboratively research on advancing their collective semiconductor manufacturing capabilities. The pact will bring researchers at the respective labs of Intel in Hillsboro, Oregon, and IBM in Albany, New York. Both companies seek to:
· Make the U.S. semiconductor industry more competitive.
· Support key U.S. government initiatives.
· Collaborate on advancing key capabilities such as hybrid cloud, AI, 5G, the intelligent edge and quantum computing.
Gelsinger offered an overview of Intel’s client, server and infrastructure processors in the pipeline.
· Tiger Lake: Ultra mobile laptops with Intel’s 11th Gen Intel Core processors with Intel Iris Xe graphics started rolling out late last year with a broader portfolio now appearing in the channel. Also on this year’s agenda are a broader lineup for business, education and desktops, introduced at January’s Consumer Electronics Show (CES).
· Alder Lake: Also at CES, 11th Gen Intel Core processors for desktops are set to appear in the second half of this year. Intel says it will provide more scalability and higher-efficiency cores. It will also be Intel’s first CPU built on an the company’s new 10nm SuperFin, designed to provide faster and more efficient desktop and mobile computing.
· Rocket Lake: A crop of 11th Gen Intel Core S-series processors for desktops and gamers are set to appear imminently.
· Ice Lake: On April 6, Intel will launch its 3rd Gen Xeon Scalable processors for data centers, 5G networks and intelligent edge infrastructure. It comes just three weeks after rival AMD introduced its Gen 3 EPYC processors. Gelsinger said early testers include the Korea Meteorological Administration and Baidu for its AI cloud servers.
· Sapphire Rapids: Early access customers have begun testing the successor to Ice Lake, Sapphire Rapids. Gelsinger said manufacturing is set to begin by year’s end with production ramping during the first half of 2022.
· Meteor Lake and Granite Rapids: The aforementioned shift to 7 nanometers begins in 2023 with the new Meteor Lake for client and Grand Rapids for data center. Gelsinger said both Meteor lake and Grand Rapids will have compute tiles and incorporate the company’s new packaging capabilities.
For the first quarter, Gelsinger announced that Intel expects exceed its prior top- and bottom-line guidance, driven by continued strong notebook demand. Demand for PC products appears strong throughout 2021.
The updated forecast reflects the industry-wide shortage of critical third-party components such as substrates, Gelsinger said.
“We are working aggressively with our supply chain partners, as well as leveraging our own unique manufacturing capabilities to solve for component shortages,” he said.
Intel issued the updated forecast in an 8K filing with the SEC on March 23.
Four years after discontinuing the popular Intel Developer Forum (IDF), Gelsinger it will return this fall in a new form. Gelsinger recalled speaking at the first IDF 24 years ago.
“It was the place for lovers of technology like me to gather and geek out about the future,” he said.
The company is resurrecting the “spirit” of IDF with the launch of Intel On. It is scheduled to take place as a live industry event this October in San Francisco.
Four years after discontinuing the popular Intel Developer Forum (IDF), Gelsinger it will return this fall in a new form. Gelsinger recalled speaking at the first IDF 24 years ago.
“It was the place for lovers of technology like me to gather and geek out about the future,” he said.
The company is resurrecting the “spirit” of IDF with the launch of Intel On. It is scheduled to take place as a live industry event this October in San Francisco.
The Intel IDM 2.0 strategy revealed by new company CEO Pat Gelsinger is an ambitious plan to restore the company’s leadership in semiconductor production. Gelsinger described IDM 2.0 as the second generation of Intel’s integrated device manufacturing model.
Intel’s Pat Gelsinger
IDM 2.0 is a broad strategy that doubles down on research and development. It expands manufacturing capacity and opens the door to working more closely with peers in the semiconductor industry. The move comes two months after Gelsinger returned to Intel as CEO after leading VMware for eight years.
During his one-hour presentation, “Engineering the Future,” Gelsinger methodically outlined key initiatives that form the basis of Intel IDM 2.0. In the slideshow above, we unpack nine key plans Gelsinger discussed
Read more about:
VARs/SIsAbout the Author(s)
You May Also Like