Intel details foundry revenue for the first time, sees H2 upturn
Cette publication existe aussi en Français
Intel has detailed revenue from its foundry business for the first time as it sees semiconductor markets recover in the second half of 2024.
“This quarter, we officially transitioned to our new operating model and introduced Intel Products and Intel Foundry. Today, for the first time, we are reporting our results to reflect the new way in which we are running the company,” said Pat Gelsinger, CEO of Intel.
Intel Foundry revenue was $4.4 billion, down 10% year on year on lower backend services and samples revenue along with lower tool sales. In addition, wafer volume was only modestly higher in the quarter with ASPs modestly down driven by pricing for mature nodes.
- Intel puts on a brave face as data centre revenue plummets
- Intel looks to 14A process in 2026
- Intel Fellow discusses high-NA EUV lithography progress
However operating profit declined by approximately $100m YoY with lower revenue being partially offset by improved factory utilization. Operating margin declined significantly QoQ (quarter over quarter) driven by higher startup costs and the conclusion of the traditional packaging business, impacting revenue.
“The Foundry P&L will remain challenged through the year, and we expect operating margins to trough in 2024 as startup costs associated with 5N4Y peak and the P&L absorbs an expected increase of roughly $2 billion in depreciation. Beyond 2024 [with] the return of compute tiles to internal process nodes, we expect to see rapid profitability improvement,” said David Zinsner, chief financial officer at Intel.
“Semiconductors are the currency that will drive the global economy for decades to come. We are one of two, maybe three, companies in the world that can continue to enable next-generation chip technologies and the only one that has Western capacity and R&D, and we will participate in the entire AI market,” said Gelsinger.
“We are reporting our results to reflect the new way in which we are running the company. Separating the internal financial reporting between Intel Foundry and Intel Products was a critical step needed to provide transparency, accountability and the proper incentives to allow both groups to make better decisions to optimize their own cost structures,” he added
“We knew that the “Day One” P&L (profit-and-loss statement) for Intel Foundry was going to spark debate, but we also knew it was important to establish a baseline and provide a target based on reasonable-to-conservative revenue and cost assumptions that we have a high degree of confidence we will achieve.”
He sees the market picking up in the second half of the year as shipments of AI chips increase.
“While 1H trends are modestly weaker than we originally anticipated, they are consistent with what others have said and also reflect some of our own near-term supply constraints,” he said.
“We continue to see Q1 as the bottom, and we expect sequential revenue growth to strengthen throughout the year and into 2025, underpinned by the beginnings of an enterprise refresh cycle and growing momentum for AI PCs, a data centre recovery with a return to more normal CPU (central processing unit) buying patterns and ramping of our accelerator offerings, and cyclical recoveries in NEX (Network and Edge Group), MBLY (Mobileye) and Altera,” said Gelsinger.
Gelsinger talks of the ‘Grovian’ culture in tribute to former Intel CEO Andy Grove, who famously said ‘only the paranoid survive.’ Intel says it is on track to launch five process technology nodes in four years, or 5N4Y.
“We have rebuilt our Grovian culture and execution engine and are on-track to completing our 5N4Y goal, which many of our stakeholders thought impossible at inception. In so doing, we are in a unique position with at-scale EUV technology, Western-based capacity and, at the very least, a level-playing field with the market leader.
Intel 20A, which helps pave the way for Intel 18A, begins production ramp in the second half of this year with Arrow Lake. We expect to release the 1.0 PDK (process design kit) for Intel 18A this quarter. Furthermore, our lead products – Clearwater Forest and Panther Lake – are already in fab, and we expect to begin production ramp of Intel 18A and these products in the 1H of ’25 for product releases in the middle of next year.”
“Establishing a foundry relationship between our products group and our manufacturing group was a critical step to achieving better structural cost.
The company has 18A processors in the fab and expects to ramp the process in 2025. “Our first Intel 18A product, Clearwater Forest, is slated to launch next year and will allow us to accelerate share gains,” said Gelsinger.
- Faraday plans 64 core ARM SoC on Intel 18A
- Intel works with 7 out of top 10 fabless companies
- Ericsson to boost 18A process
He cites the building of leading fabs in the US as a key competitive advantage, although competitors TSMC and Samsung also have CHIPS Act funding to build 2nm fabs in Arizona and Texas. Rapidus is also building a 2nm fab with back end processing in Japan and looking at more aggressive process technology.
“More competitive wafers, combined with our position as the only company manufacturing with leading-edge wafers outside of Asia is drawing strong interest from potential external customers. It is important to note that our leadership in advanced packaging creates more value in our wafer technologies, and wafer-level assembly and base die opportunities further fill our factories and extend the useful life of our tools for increased financial returns,” he said.
The foundry market is expected to grow from $110 billion today to $240 billion by 2030, with almost 90% of the growth coming from EUV nodes and advanced packaging.
“Given this backdrop, we have clear line of sight to becoming the largest system foundry for the AI era and the second largest overall by 2030, building on our EUV/High NA process technology, leadership in advanced packaging, manufacturing capacity, our systems expertise and the surge in AI demand,” he said.
The company spent $5n on fabs and capex in the quarter, with operating cash flow negative at $1.2 billion.
Intel 18A
The company has six customers for its Intel 18A process used by Intel Foundry and has processed 20 test chips on the technology. This week Intel Foundry received funding from the US Department of Defense (DoD) for Phase Three of its Rapid Assured Microelectronics Prototypes – Commercial (RAMP-C) programme.
This will see the tape-out and testing of early defence industrial base (DIB) product prototypes and RAMP-C customers can now begin manufacturing commercial and DIB product prototypes on the 18A process technology. These include Microsoft, Nvidia, IBM, Boeing and Northrop Grumman.
“This marks another significant moment in advancing our collaboration with DoD on this program. For the first time in decades, US government and defense industrial base customers will have access to one of the industry’s leading-edge process technologies at the same time as commercial customers,” said Kapil Wadhera, Intel vice president for Foundry Services and general manager of the Government Engagements and Business Operations Group.
“Intel Foundry, a performer on the RAMP-C project, has met their milestones and metrics for Phase Two of the project and has recently been awarded funding to start their Phase Three effort. R&E’s program, managed through Navy Crane and executed through an OTA with NSTXL, is ensuring that the DoD stays at the cutting edge of microelectronics technologies while facilitating close collaboration between the defense industrial base and commercial companies. RAMP-C intends to demonstrate prototype production of Intel 18A chips in 2025 to deliver unprecedented processing performance for the DoD,” said Dr. Dev Shenoy, secretary of defense for Research and Engineering, principal director for Microelectronics.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
