Will Intel surpass TSMC this year?

Latest update time：2024-02-20

Reads：

????If you want to see each other often, please mark it as a star???? and add it to your collection~

Source : Content compiled from eejournal by Semiconductor Industry Observation (ID: ic bank ) , thank you.

According to Intel's previously announced plans, the company is expected to catch up with and surpass TSMC, the current recognized leader in semiconductor processes, by 2024 or 2025. This is definitely an achievable goal as Intel rolls out its 10nm SuperFin (now called Intel 7) and 7nm (now called Intel 4) process nodes.

In this article, we will focus on this issue.

First, we'll focus on the what and when: planned semiconductor process nodes and when they will occur. Then we'll focus on where: where Intel is investing in new fabs and fab upgrades. Next, we'll talk about who, as Intel has announced some interesting foundry partners. Finally, we'll focus on why. The takeaway: Intel views process node leadership as a survival aspect of its business.

Intel is betting that its 10nm SuperFin node won't require EUV lithography. It was a conservative gamble that didn't pay off. The rollout of the 10nm node was delayed and Intel began to lose its race against the Moore's Law curve. Another result is that Intel did not gain EUV experience entering the 7nm node, so the node was late as Intel's process engineers needed to adapt to the EUV learning curve. The two process nodes are now called Intel 7 and Intel 4, and both are in production.

Shortly after Pat Gelsinger took over as Intel CEO in early 2021, he announced process node plans through 2025. These plans call for conquering five process nodes over four years, including the newly named Intel 7 and Intel 4 nodes. Intel announced in mid-2023 that the intel 3 process node (the company's last generation of FinFET nodes) has reached yield and performance targets and will enter mass production in 2024 for the company's Granite Rapids and Sierra Forest CPUs.

The next process node, Intel 20A (the "A" stands for Angstrom), introduces two major changes to the company's process technology.

First, FinFETs are out and RibbonFETs are in. RibbonFET is Intel's name for a gate-all-around (GAA) FET that surrounds the FET channel with a gate to improve control of gate current and improve performance. Many companies are developing GAA FET processes. In mid-2023, Samsung announced that it had begun initial manufacturing of its 3nm process technology, which is based on the company's multi-bridge channel FET (MBCFET). This is Samsung's version of GAA FET.

Intel 20A also introduces Intel's backside power technology, PowerVia, to the manufacturing process node. PowerVia technology introduces a new set of process steps in IC manufacturing as it moves the entire power delivery system from the top of the chip (where it has been since Fairchild Semiconductor's Jean Hoerni patented a planar process for IC manufacturing in 1959) to the chip The back. Powering transistors fabricated on the front of the chip from the back requires drilling countless microscopic holes in a thinned silicon wafer and then filling the holes with metal to make connections from the backside power delivery system to the transistors.

Backside powering has many advantages, but two of the most important are strengthening the metal used to power the chip's transistors, thereby reducing on-chip power drops and noise, and eliminating power distribution routing from the front of the chip, freeing up metal on the top of the chip layer space, thereby greatly reducing the restrictions on signal wiring.

Both RibbonFET and PowerVia technologies have risks. Typically, process development teams only introduce one major change at a time. However, Gelsinger is impatient and is pushing for a forced march on the process node so both technologies will appear on the Intel 20A process.

To mitigate some of the risk, Intel process engineers created an internal-only process node that added the PowerVia backside power delivery system to the Intel 4 process node. Intel reports that this change will increase performance by 6% in mid-2023 alone. Intel claims that the Intel 20A process will enter production in the first half of this year. At the end of 2023, Intel CEO Kissinger announced that the subsequent process node Intel 18A will be put into production before the end of 2024. TSMC has stated that its equivalent process node N2 will be in production in the second half of 2025, so it seems that if everything goes according to plan, Intel may indeed be in the lead by the end of this year.

Developing new process nodes is one thing. Producing chips in volume using a new process node is another story, so Intel has also embarked on a huge, capital-intensive project to upgrade some of its existing fabs and build entirely new ones. The construction boom will require tens of billions of dollars in commitments from Intel and interested governments.

Intel upgraded a manufacturing facility in Leixlip, Ireland and recently enabled Intel 4 process at the location. Intel is building two new factories in Chandler, Arizona to run the Intel 20A process. The company has established a new manufacturing site and is building an Intel 18A fab in Licking County, Ohio. Intel also recently announced plans to build a new wafer fab in Magdeburg, Germany, to manufacture chips using a process node inherited from Intel's 18A. The company is also building a factory in Kiryat Gat, Israel, to make chips using its older FinFET process node, and it has announced plans to build new packaging facilities in Malaysia and Poland.

(Note: With so many fab construction and improvement projects underway and being announced, this long list is undoubtedly incomplete or will be completed soon. Additionally, Intel recently announced that due to the slow government rollout With funding from the CHIPS Act, construction of the Ohio plant will be delayed. With so many projects underway, more such delays are possible, if not unlikely.)

There’s a long-standing adage that applies to the semiconductor industry: “Fill the Fab.” This is the only way to control manufacturing costs. Idle fabs burn a lot of money. Intel has been lucky to have long had much-needed products to meet the needs of its fabs, but Gelsinger is a veteran of the semiconductor industry and knows it takes a variety of chips to meet the needs of the fabs. That's one of the reasons he created Intel Foundry Services (IFS). As a result, Intel has been striking foundry deals with customers to help maintain capacity at its fabs.

In addition to direct foundry customers, Intel has also been striking deals with other semiconductor foundries. When Intel's deal to acquire Tower Semiconductor was blocked in August 2023, the two companies announced a foundry deal the following month. Intel will provide foundry services and 300mm manufacturing capabilities for Tower at its facility in Rio Rancho, New Mexico. As part of the deal, Tower will purchase up to $300 million worth of manufacturing equipment for the New Mexico facility to run its proprietary semiconductor process. Tower's semiconductor processes are not at the forefront of lithography technology, but they are at the forefront of applications such as power management, RF signal processing and low-power operation. This month, Intel and UMC announced a similar collaboration to develop the 12nm FinFET process node and produce it at Factories 12, 22 and 32 of Intel's Ocotillo Technology manufacturing facility in Chandler, Arizona.

Robert Noyce and Gordon Moore founded Intel in 1968. By 1971, the company had developed the first commercial DRAM, thus creating the DRAM market, the first commercial microprocessor, thus creating the microprocessor market, and the first EPROM, thus creating the non-volatile semiconductor memory market. Intel has long been among the top ten semiconductor manufacturers. The company took the No. 1 spot in semiconductor sales in 1992 and maintained that lead until 2018, when it became the largest semiconductor seller due to rising DRAM prices. However, DRAM prices fell and Intel was once again in the lead.

When products in these markets lose high profitability, Intel will abandon the entire market.

In 1985, Intel abandoned the DRAM market to prevent Japanese DRAM suppliers from eating into Intel's market share, thus saving the company from disappearing. Intel has been in and out of the non-volatile semiconductor memory field. It has currently exited the non-volatile semiconductor memory field and sold its flash memory business in 2020. In addition, Intel is currently spinning off the FPGA business it acquired when it acquired Altera in 2015. But it doesn't matter, because the spin-off FPGA company will soon become a major customer of IFS.

However, Intel cannot abandon the core product lines of its microprocessor business or disappear because it has no other large business to rely on. To retain this business, Intel must fend off its long-time rival AMD, which has been making inroads into the PC and server CPU business, in part because Intel has lost its process leadership. So, this is an existential question for Intel. Since TSMC produces AMD's processors, Intel's strategy for leading in microprocessors must include beating TSMC at every new semiconductor process node while designing advanced processors for each new node. Losing this game would mean Intel is finally forgotten.

Gelsinger's push through five process nodes in four years is part of the strategy to win this race, and I think Gelsinger's IFS and its planned role as a major semiconductor foundry is part of his strategy to keep Intel technology development and manufacturing teams One of the ways. This is not to say that there were not countless other elements in Kissinger's strategy. Taking Gelsinger and TSMC at their word, Intel does seem to be able to regain semiconductor process leadership from TSMC this year. If Gelsinger and Intel do succeed, it would be another hard-won victory for a company that has contributed so much to the semiconductor industry over the past half century.

Original link

https://www.eejournal.com/article/will-2024-really-be-intels-year-to-retake-the-lead-in-semiconductor-process-technology/

Click here to follow and lock in more original content

END

*Disclaimer: This article is original by the author. The content of the article is the personal opinion of the author. The reprinting by Semiconductor Industry Watch is only to convey a different point of view. It does not mean that Semiconductor Industry Watch agrees or supports the view. If you have any objections, please contact Semiconductor Industry Watch.

Today is the 3681st issue shared by "Semiconductor Industry Observation" with you. Welcome to pay attention.