Track Hyper | Intel "considers" halting the promotion of 18A process technology

Author: Zhou Yuan / Wall Street Insight

On July 2, industry news reported that Intel is making another strategic adjustment: Intel CEO Lip-Bu Tan is "considering" adjusting its foundry business—stopping the promotion of the Intel 18A and subsequent iteration Intel 18A-P process technology to foundry customers, and focusing on tackling the Intel 14A process.

On July 3, Wall Street Insight confirmed this matter from authoritative sources, but whether Intel will actually abandon the promotion of the Intel 18A technology node is expected to have a conclusion only in September this year.

This is not the first time Intel has adopted a "leapfrog" strategy in its foundry technology roadmap.

In September 2024, Intel announced it would "skip the productization" of the Intel 20A node, redirecting engineering resources from Intel 20A to Intel 18A to reduce capital expenditures, with plans to launch Intel 18A in 2025.

It should be noted that this is not an isolated strategic decision but a strategic contraction of this semiconductor giant after encountering real bottlenecks in the advanced process race, and its impact will be reflected in multiple aspects.

Intel's investment in the 18A process is likely to face considerable financial pressure. However, there are currently no official reports detailing Intel's specific investment amount in the 18A process, so indirect speculation can only be made.

According to Intel's 2024 financial report, the foundry division (IF) recorded a net loss of $13.4 billion for the year, accounting for 71.28% of Intel's total annual loss of $18.8 billion.

Industry analysts estimate that the R&D investment in 18A and related technologies has already cost several billion dollars. If Intel stops promoting the 18A process to external customers, it may face asset impairment processing in the range of hundreds of millions to several billion dollars.

Originally, Intel planned to digest costs through a dual-track model of "own products + external foundry," but currently, the only external customers for 18A are Amazon and Microsoft; among them, the former has a longer cooperation cycle, bringing estimated revenues of several billion dollars, which is difficult to scale in the short term and effectively reduce costs.

What is the competitive landscape of the global foundry market?

According to a research report released by Counterpoint Research on June 24 this year, the global Foundry 2.0 market revenue is expected to reach $72 billion in the first quarter of 2025, a year-on-year increase of 13%.

The concept of "Foundry 2.0" was proposed by TSMC at its second-quarter earnings call in July 2024, mainly including packaging, testing, photomask production, and others, as well as all integrated device manufacturers (IDMs) excluding memory chip manufacturers. This differs from the traditional concept of "foundry" limited to "wafer manufacturing production foundry."

Therefore, even Intel, leveraging its own processor manufacturing and advanced packaging business, has become the second-largest Foundry 2.0 manufacturer globally, with a market share of 6.5%, an increase of 0.6 percentage points quarter-on-quarter, but a decline of 0.3 percentage points year-on-year Currently, there is a lack of global revenue market share rankings for traditional wafer foundries for the entire year of 2024 and the first quarter of this year, making it impossible to determine the extent of financial improvement Intel would experience if it stops promoting 18A to customers.

However, after Chen Liwu took office in March this year, he immediately announced a cost-cutting plan in April, leading the market to speculate that it would involve large-scale layoffs and the cancellation of certain projects.

Currently, the former has been fully validated by the market, while the latter, according to industry information, shows that in June this year, Chen Liwu shared with colleagues that the attractiveness of the 18A process to external customers is declining. Therefore, Chen Liwu believes it is wise for Intel to stop offering 18A and its performance-enhanced version 18A-P to foundry customers.

But this does not mean Intel will completely abandon the 18A series process; although it will stop external promotion, Intel is likely to still apply it internally.

According to previous plans, Intel 18A will be used to produce the newly announced next-generation client processor Panther Lake, as well as the Clearwater Forest processor aimed at data centers.

At the same time, Intel will continue to manufacture new products based on the Intel 18A process for customers with whom it has already committed to cooperate, such as possibly producing a small number of chips for Amazon and Microsoft.

It is reported that Chen Liwu suggested allocating more resources to the completion and promotion of Intel's next-generation node 14A. According to the original plan, this node is expected to enter risk production in 2027 and achieve mass production in 2028.

According to technical data disclosed by Intel, Intel 14A offers a 15%-20% performance improvement per watt compared to Intel 18A, a 30% increase in density, and a 25%-35% reduction in power consumption.

At the same time, both the Intel 14A process and its evolved version 14A-E will adopt the second-generation RibbonFET gate-all-around (GAA) transistor technology, as well as the new Power Direct direct contact power supply technology (18A uses PowerVia back power supply technology), with production technology utilizing High NA EUV.

Chen Liwu's confidence in Intel 14A comes from the cost advantages brought by the use of High NA EUV technology.

The transistor spacing formed by a single exposure of High NA EUV is equivalent to three exposures of conventional Low NA EUV, which requires about 40 steps to produce the pattern. Clearly, the former is more efficient, thus effectively reducing costs. This makes the Intel 14A process potentially more advantageous than TSMC's A14 process.

Despite these benefits, whether Intel 14A (including the evolved version 14A-E) can successfully replace 18A as the main technology service for IF and obtain board approval remains uncertain, as this involves the scale of R&D investment and the efficiency of technology implementation, which are currently unclear. The successful commercialization of a technology depends on the overall efficiency of an organization.

It is possible that by September, the Intel board may decide to continue promoting 18A's external services. One thing that is currently clear is that the plan to launch the Panther Lake laptop chips using the Intel 18A process remains unchanged for the second half of this year The capital market's reaction to Chen Liwu's "self-inflicted injury" is: as of the close of the US stock market on July 2, Intel's stock price fell by 4.25%.

Overall, considering the possibility of stopping the promotion of 18A, breaking the deadlock requires a systemic transformation that goes beyond a single process. Therefore, as of now, Chen Liwu is still addressing the systemic challenges faced by Intel