Intel Q1 2025 conference call: Full-year capital expenditure target lowered to $18 billion

According to Zhitong Finance APP, recently, Intel (INTC.US) stated in its Q1 2025 earnings call that it expects to keep capital expenditures at $18 billion. The company will adopt a more aggressive approach to improve the return on invested capital, thereby reducing capital expenditures. This situation may continue next year, but it is too early to discuss guidance for capital expenditures in 2026, and next year will also utilize assets under construction. In the long term, this is a capital-intensive model, with a capital intensity of about 25% under a complete integrated device manufacturer (IDM) model, which is a reasonable long-term assumption.

In addition, regarding the development of the data center business, Intel expressed optimism for the remainder of this year. The product roadmap includes the launch of Granite Rapids and the Xeon 6 series, which will drive traditional business integration. Intel is also one of the choices for Co-Packaged Optics (CPO) in the artificial intelligence field, and improvements have begun in the telecommunications sector. However, there are still significant macro concerns. In the data center, similar to client business, there is strong demand for legacy products, and related issues are being addressed from a supply perspective.

Q&A

Q: How to balance Intel's internal product roadmap with foundry business?

A: On one hand, we need to flatten the organizational structure, focus on the right products, and deliver customer solutions; in terms of products and roadmaps, focus on key products to ensure timely delivery, performance, and power to meet customer needs. For the foundry business, 18A is important for Panther Lake, and we need to gradually improve yield and reliability to enable internal customers to use it. On the product side, we should strive to drive product performance based on internal and external conditions.

Q: What are Intel's gross margin targets and influencing factors for 2025 and even 2026?

A: In 2025, the client product mix will shift more towards Lunar Lake, which faces margin pressure due to memory and packaging issues, impacting gross margins; the AT&A business (advanced packaging and testing) starting in Arizona has related costs, which will also put pressure on gross margins for the year. The situation will improve in 2026, with Panther Lake's production increasing next year, and its margin is better than Lunar Lake. Additionally, using 18A for Panther Lake will bring the benefit of wafers returning to our own factories for production, resulting in a combined margin enhancement effect. However, gross margins are ultimately influenced by revenue, and tariffs may be unstable this year. Although the company has a global supply chain to mitigate some headwinds, optimizing the network takes time.

Q: What is Intel's actual gross margin for Q1, and will Q2 gross margin be below guidance?

A: Q1 gross margin was better than expected, putting relative pressure on Q2, and changes in tariffs will alter the gross margin expectations for the year. Regarding Q1 gross margin, the biggest positive factor was revenue exceeding expectations, with Raptor Lake accounting for a larger share of product sales than Lunar Lake, which is beneficial for gross margins. Additionally, management expenses were also well managedIf these two factors are excluded, it will be closer to the guidance.

Q: What is Intel's policy regarding the announcement of new third-party foundry customers, and what is the current progress? Will there be an increase in tier-one customers this year?

A: Intel's foundry business's top priority is to increase the business volume of internal customers (such as the previously mentioned Panther Lake). The next step is to build trust with foundry customers and ensure the robustness of the PDK (Process Design Kit) and timelines. In terms of process technology, the focus is on improving yield and reliability, adopting a customer service mindset, and matching different customers' design tools and methodology styles to optimize solutions. After that, efficiency will continue to be improved to scale the business.

Q: How do you evaluate the Panther Lake and Clearwater Forest in the AA product (advanced packaging) portfolio? Do they belong to leading products, and what work remains to achieve a leading position?

A: The company's top priority is to focus on product execution, delivering on time, and meeting customer and industry trends and workload demands. For Panther Lake, the 18A process is very important, with the first products launching by the end of this year and more SKUs next year. This product is competitive in the market and can attract significant customer interest. Regarding Clearwater Forest, it has unique packaging technology, and the company has decided to prioritize the launch of Panther Lake, with Clearwater Forest set to launch in the first half of 2026. It is based on E-core technology and is a derivative of Granite Rapids, leaning more towards specific-use products rather than leading performance products in the market. Currently, both products are being delivered as planned.

Q: What is the reason for the limited capacity of Intel 7 in the foreseeable future, and what impact will it have?

A: On one hand, the Intel 7 portion of Raptor Lake and Raptor Lakes is performing well, with capacity planning nearing its limits. Changes in capacity allocation have led to restrictions, but this is generally good for the factories, as they will utilize the network to produce more wafers under constrained conditions, reducing unit costs. On the other hand, influenced by macroeconomic factors and the overall economy, there is strong demand in the consumer market for n - 1 and n - 2 products due to relatively stable prices; in the commercial sector, there is demand for Windows 10 updates and strong demand for AIPC. Although the slowdown in AIPC growth is actually beneficial for the company's gross margin, the company will continue to invest in AI PCs, balancing the product mix and capacity across all fabs.

Q: Is the company's AI strategy to reshape x86 for success in the AI field, or to adopt a broader product portfolio including ARM? Is there a greater focus on edge AI?

A: The company's AI strategy is to first understand the workloads, ensure effective delivery, and collaborate with industry leaders to create dedicated silicon chips and software for specific platforms, becoming a computing platform. Edge and inference areas are important directions, and the company is exploring new architectures and disruptive platformsQ: How will the data center business develop in the remaining time of this year, especially in the second half?

A: In the first quarter, the data center business achieved growth above expectations, mainly driven by several hyperscale customers. We are optimistic about the remaining time this year, as the product roadmap includes the launch of Granite Rapids and the Xeon 6 series, which will drive traditional business integration. Additionally, Intel is one of the choices for Co-Packaged Optics (CPO) in the artificial intelligence field, and there have been improvements in the telecommunications sector. However, there are still significant macro concerns. In the data center, similar to client business, there is strong demand for legacy products, and we are addressing related issues from a supply perspective. The main goals mentioned in last year's fourth-quarter earnings report were to stabilize market segment share, generate profits, and increase average selling prices, which will be the focus for the remainder of this year. The product portfolio is strong, with robust performance from hyperscale and enterprise customers, but market segment share in other parts of the world faces challenges, which will be a key focus.

Q: How is capital expenditure (CapEx) planned for 2025-2026, and will it be rationalized based on existing scale?

A: For 2025, we believe we can control expenditures to around $18 billion. The company will take a more aggressive approach to improve the return on invested capital, thereby reducing capital expenditures. This situation may continue next year, but it is too early to discuss guidance for 2026; next year will also utilize assets under construction. From a long-term perspective, this is a capital-intensive model, with a capital intensity of about 25% under a fully integrated device manufacturer (IDM) model, which is a reasonable long-term assumption.

Q: Why is the demand for older generation products (such as Raptor Lake) greater than for newer generation products (such as Meteor Lake and Lunar Lake)? Is it because Intel is promoting older products for higher profit margins?

A: The promotion of older products is not based on profit margins. Customers have a greater demand for N-1 and N-2 products so they can continue to offer prices that consumers truly need. Macroeconomic concerns and tariff issues have led everyone to hedge risks from an inventory perspective. Raptor Lake is an excellent product, and Meteor Lake and Lunar Lake are also good, but they not only have a higher cost structure for Intel, but also a higher average selling price for original equipment manufacturers (OEMs). From the OEM's perspective, they have already reduced the cost curve with Raptor Lake, allowing them to offer the product at a better price, so it is mainly the macroeconomic environment and overall economic conditions, as well as customers' risk hedging strategies, that have led to this situation.

Q: Considering macro events, tariffs, and the shift in demand towards older products, how will the Panther Lake product, set to launch at the end of the year, fare?

A: The launch timing of Panther Lake aligns with the launch schedules of Meteor Lake and Lunar Lake, catering to customers' preferences for bringing products to marketFrom the perspective of performance and price, it is an excellent product for customers, and strong market acceptance is expected. There is still very strong demand for AI PCs in the commercial sector, as companies want products with AI capabilities and future readiness when deploying and upgrading equipment, so the demand in the commercial sector will not change. Typically, the promotion of such products sees rapid growth in the commercial sector first, followed by the consumer market. Although the economic situation at the end of the year needs to be considered, there is great confidence in the Panther Lake product and customer feedback.

Q: Why is the performance guidance for the second quarter relatively weak, and what is Intel's market share situation in the server market for the next few quarters?

A: In the first quarter, several hyperscale customers performed strongly, but this situation is not expected to continue into the second quarter. From an overall macro demand perspective, there are concerns about tariffs and restrictions, and the demand for older generation products exceeds current supply, requiring balance. Intel will strive to stabilize its market segment share, but this will also bring a lot of pressure on margins and average selling prices (ASP), requiring balance throughout the year. Granite Rapids is a great product, and customers are very interested in it, but there is also competition in the market. Intel will cautiously plan for the year and commit to a performance target that can be surpassed.

Q: Is Intel's goal of achieving a 70% internal production ratio for chips by 18A in the Panther Lake project still unchanged? Has a decision been made regarding internal versus external production for Nova Lake, and what is the ratio?

A: The goal of achieving approximately 70% internal production ratio for chips by 18A remains unchanged, and Panther Lake will help achieve this goal. For Nova Lake, the process nodes to be used have been optimized at the SKU level, and products will be produced at TSMC and Intel's internal ULC. Overall, the wafers produced on Intel's process for Nova Lake will be more than those for Panther Lake, continuing the commitment to drive wafer growth with internal foundry partners. At the same time, Intel will continue to balance internal and external wafer manufacturing, which is part of the capital strategy to maintain reasonable capital intensity and avoid over-investing capital or achieving poor results in asset returns.

Q: In the first twenty days of this quarter, was there any early procurement situation in client and data center businesses, geographic regional dynamics, and different customer behavior?

A: The first twenty days of this quarter started relatively strong, but macro factors are expected to be affected by tariffs, and the next quarter may be weaker than this quarter's strong start. However, it cannot be completely ruled out that this quarter may remain strong overall; if so, performance may be at the high end of guidance, but the impact of tariffs on the macro situation has been incorporated into expectations.

Q: What is Intel's future strategy for using products like Gaudi and Falcon Thors in data centers, and are there plans to launch new products?A: In the first five weeks of Lip-Bu's tenure, he and Sachin spent a significant amount of time re-evaluating Intel's artificial intelligence strategy, asset portfolio, and how to enter the market to compete. Intel has adopted a workload-first approach and decided in Q4 not to launch Falcon Shores, continuing to advance POR Jaguar Shore, with Jaguar shores still on the product roadmap. In the coming months, Sachin and Lip-Bu will publicly announce the AI roadmap and plans during this period. As they engage in in-depth discussions with customers about the workload roadmap, Intel will adopt an aggressive strategy, as this segment of the data center market has not yet entered competition and requires a strong product portfolio to meet customer demand for alternatives.

Q: How long will it take for Intel's business to improve in reality, what are the appropriate metrics to measure progress, what are the optimization metrics for the company, and what is the timeline for achieving the goals?

A: Currently, there is no quick fix. The company is advancing its roadmap, rapidly updating with the team, and clarifying new workloads in CPU, GPU, and AI. It will launch both short-term and long-term products, potentially adopting some disruptive technologies in the short term and collaborating with relevant parties to bring them to market faster to meet customer needs, which requires continuous attention. In terms of metrics, the company aims to have the best products, especially for edge computing, with high energy efficiency and ideal performance. Timely product launches are also critical, and the company is working towards meeting industry and customer demands.

Q: Is the progress of the 18A technology and single-digit revenue growth sufficient to make Intel's IDM structure business viable, or does it need to rethink the IDM structure?

A: The company hopes to take a balanced approach, ensuring that the foundry can meet internal customer needs and serve Intel's customers well, ensuring ease of product delivery and timeliness. At the same time, the company views TSMC as a partner, and both parties are looking for areas of collaboration to achieve a win-win situation. Additionally, the company will continue to improve the yield efficiency of AT&A, focusing on 14A technology in line with customer demand and effectively utilizing existing manufacturing capacity.

Q: Do the operating expense structures for 2025 and 2026 include costs related to the divestiture of 51% of Altera?

A: The $17 billion operating expense for 2025 and the $16 billion for 2026 actually include Altera's operating expenses. Once the divestiture is completed, the target of $16 billion will be reduced accordingly based on Altera's operating expense expenditures. However, since the transaction has not yet been completed, it is difficult to consider its impact in the forecast, so it is temporarily included.

Q: How do you view rack-scale networks, and what elements within Intel can compete in the construction of cloud infrastructure AI? What strategic building blocks are still needed to drive the implementation of related strategies?A: A comprehensive layout is needed in both hardware and software, fully utilizing XPU, CPU, and GPU to drive progress. The IPU product (intelligent network card) has achieved great success, and its revenue is expected to double in 2024-2025. Optics is also a key element of rack-scale architecture, and Intel is the only foundry that offers customers options for optical module foundry services. In addition, customers favor the x86 ecosystem and its surrounding software; if they can build AI infrastructure using x86, they will be very interested. The company already has a large custom design order, and more are expected in the future