Intel Q1 Earnings Preview: Triple Test of CPU, 18A Yields and Foundry Orders

On April 8, the chief analyst at SemiAnalysis pointed out that CPUs are facing a severe capacity shortage. Currently, the ratio of CPUs to GPUs in AI data centers is roughly between 1:4 and 1:8; however, driven by the widespread adoption of AI agents, this ratio is expected to quickly shift toward 1:1 or 1:2.

Forecast data indicates that the data center CPU market size will grow from $25 billion in 2026 to $60 billion by 2030, and with the incremental growth brought by AI agents, it could potentially reach the $100 billion level. The supply side is also under pressure, with Intel's lead times for Chinese customers extended to six months and price quotes in the China region rising by more than 10%. Both Intel and AMD raised prices for certain product lines at the end of the first quarter.

The company also noted that current capacity cannot fully meet strong demand, with supply bottlenecks expected to peak in the first quarter and begin easing gradually in the second quarter. However, memory chip shortages have constrained PC assembly progress, and full-year ODM shipments are projected to see a double-digit percentage decline. Revenue for the CCG department may fall by more than 13% quarter-over-quarter, which would partially offset the growth momentum from the server business.

Regarding the earnings report, the market wants to see whether the data center business is growing fast enough and its scale is large enough to completely offset the decline in the PC business and lift the overall revenue curve.

If the supply-demand dynamics of server CPUs determine current sales volume, the mass production curve of the 18A process points to a more fundamental question: How much does Intel actually earn for every chip it sells?

The 18A process has reached the high-volume manufacturing stage with yields maintained above 60%, which is the primary condition for commercial profitability. Lip-Bu Tan noted that yields were suboptimal when he took over, but after introducing professional partners such as PDF Solutions and KLA, yields have been rising steadily at a rate of seven to eight percentage points per month.

KeyBanc pointed out that while current yields are lower than the 70% to 80% levels seen during the early mass production of TSMC's 2nm process, they outperform Samsung's SF2, which remains below 40%. The 18A process utilizes RibbonFET gate-all-around transistors and PowerVia backside power delivery as two key technologies, increasing performance per watt by 15%, chip density by 30%, and reducing overall power consumption by approximately 25%.

Panther Lake is the first consumer product to utilize the 18A process, making its official debut at CES 2026 with a total platform computing power of 180 TOPS.

CFO David Zinsner previously stated clearly that while current yields can support shipment volumes, they cannot support normal margins; he expects to reach target cost levels by the end of 2026 and align with industry averages by 2027.

Whether gross margins return to normal levels following the mass production of 18A will serve as a key indicator of whether Intel has truly returned to a profitable trajectory.

Intel's foundry business is the most promising chapter of its transformation story. A UBS research report indicates that Intel is currently at a critical juncture for securing multiple major contracts, with the release of the 14A Process Design Kit (PDK) version 1.0 seen as a pivotal milestone.

At this point, Amazon ( AMZN) has already revealed its hand, officially signing a multi-year, multi-billion-dollar custom chip agreement with Intel to use the 18A process for manufacturing AI fabric chips. This represents the most significant substantive order for Intel's foundry business from an external customer to date.

Apple ( AAPL) is reportedly evaluating shifting some entry-level M-series laptop processors to Intel's 18A-P node for foundry services, with a target date of 2027. Google ( GOOGL) is also considering adopting Intel's EMIB and Foveros advanced packaging technologies for some of its TPU product lines.

Even more symbolic is Terafab. On April 7, Intel announced its participation in Elon Musk's Terafab superchip project, with an investment of approximately $20 billion to $25 billion during the pilot phase and a long-term goal of one terawatt of annual computing power. Regardless of the project's ultimate outcome, it adds a layer of 'anchoring effect' to the narrative of Intel's foundry business at this stage.

If a new substantial foundry order is announced during the Q1 earnings call, the valuation methodology for the business will shift from being expectation-driven to order-verified. Conversely, if the company remains vague on this matter during the call, the optimistic expectations already reflected in the current stock price will face pullback pressure.

Ahead of the earnings report, several institutions have aggressively raised their price targets. Susquehanna raised its price target from $65 to $80, maintaining a "Neutral" rating; UBS raised its target from $51 to $65, also maintaining a "Neutral" rating; TD Cowen raised its target from $50 to $60, maintaining a "Hold." Overall, institutions acknowledge Intel's strategic transformation direction but remain reserved regarding its short-term stock price performance.

The core reason for institutional caution lies in valuation; Intel's current forward P/E ratio is approximately 122x, while NVIDIA's is approximately 23x. TD Cowen noted that the current valuation largely reflects the impact of AI agents driving CPU demand, which is redefining Intel's value, rather than reflecting realized financial data.

Data from KeyBanc indicates that server CPU capacity for Intel and AMD in 2026 is virtually sold out, with lead times for certain high-performance CPU models extending to six months and general price quotes rising by about 10% during the year. However, supply bottlenecks and improvements in 18A process yields remain pressing practical issues that management must address.

Intel's current stock price already factors in significant optimistic expectations, which will be put to a critical test by the earnings report released on April 23. What the market truly needs to see is not another narrative regarding AI CPUs and foundry transformation, but hard evidence of server CPU order growth, relief of supply bottlenecks, and actual data on the 18A yield ramp-up. When expectations are pushed to the extreme, only data can deliver the final verdict.

This content was translated using AI and reviewed for clarity. It is for informational purposes only.