Intel’s second-quarter financial results have provided the most tangible evidence to date that the company’s ambitious manufacturing turnaround—a cornerstone of CEO Lip-Bu Tan’s strategy—is finally gaining genuine traction. While the semiconductor giant continues to grapple with the structural complexities of reinventing itself as a merchant foundry, the recent performance metrics suggest that the "Intel Foundry" engine is beginning to hum with newfound efficiency.
However, beneath the improved operating margins and reduced losses lies a persistent, central tension: Intel is currently its own best customer. As the company moves to prove it can support the demanding, diverse needs of external chip designers, it faces a formidable challenge in transforming its historically insular manufacturing culture into a world-class foundry service capable of competing with the likes of TSMC.
Main Facts: A Turnaround in Motion
Intel Foundry reported revenue of $5.8 billion for the second quarter, a significant climb from the $4.4 billion recorded during the same period last year. More importantly, the segment’s operating losses narrowed substantially, dropping from $3.2 billion to $2.1 billion. The operating margin also saw a marked improvement, moving from a negative 71.7% to a negative 36.2%.
According to Intel, these gains are not accidental. They are the direct result of systematic improvements in fab yield, reduced cycle times, and a significant increase in overall fab scale. By optimizing these core manufacturing pillars, Intel has effectively lowered the cost basis for its internal product lines, such as its next-generation Xeon server processors and upcoming AI PC platforms.
Yet, the composition of that revenue remains a point of scrutiny. CFO David Zinsner disclosed that of the $5.8 billion in foundry revenue, only $293 million was generated from external third-party customers. This highlights the "Intel-heavy" reliance that remains the primary hurdle for the foundry’s long-term independence and credibility as a merchant provider.
Chronology: Building the Foundry Foundation
The trajectory of Intel’s foundry ambitions has been marked by a series of technical milestones and strategic pivots:
- 2023-Early 2024: Intel accelerates its process technology roadmap, focusing on the deployment of "Intel 4" and preparing for the rollout of "Intel 18A."
- Q1-Q2 2026: Intel achieves significant operational milestones, including a 50% sequential increase in 18A output and a 50% cost reduction for its "Panther Lake" client processor architecture.
- Mid-2026: The announcement of a strategic collaboration with cybersecurity leader Fortinet signals a move toward customized ASIC (Application-Specific Integrated Circuit) development, marking a shift in how Intel engages with non-traditional silicon partners.
- The Path to 2028: Intel reaffirms its commitment to high-volume manufacturing for its "Intel 14A" process, positioning it as the next major leap in transistor performance and density.
Supporting Data: The ASIC Strategy
A critical, often overlooked aspect of Intel’s foundry strategy is its "purpose-built silicon" business. During the recent earnings call, leadership noted that this segment is approaching a $2 billion annual run rate, with projections to double that figure to $4 billion in the "not-too-distant future."
This model is distinct from the traditional "foundry-only" approach favored by TSMC. Rather than simply accepting a "GDSII" file from a customer, Intel is increasingly offering a "menu of services" that includes:
- Back-end design services: Helping customers optimize their layouts for Intel’s specific process nodes.
- Custom ASIC integration: Combining Intel’s own IP (such as high-performance processors) with customer-specific requirements.
- Advanced Packaging: Leveraging Intel’s EMIB (Embedded Multi-die Interconnect Bridge) technology to connect disparate dies into a single, high-performance package.
This strategy is clearly designed to capture the growing segment of systems companies—in sectors like automotive, cloud infrastructure, and networking—that require custom silicon but lack the massive, internal design teams found at companies like Apple or Nvidia.
The Fortinet Test Case
The recently announced partnership with Fortinet serves as a real-world stress test for Intel’s service model. Fortinet has tapped the "Intel 4" process for its next-generation SP6 security processor.
In an exclusive exchange with EE Times, an Intel spokesperson clarified that this is not merely a pilot program or an academic evaluation. "Fortinet is an Intel Foundry customer," the spokesperson stated. "They are using Intel’s custom ASIC capabilities… to develop and produce a solution tailored to their requirements."
While the commercial scale and specific production timelines remain shrouded in confidentiality, the agreement underscores Intel’s willingness to act as a design partner. Fortinet maintains control over the architecture and front-end design, while Intel handles the back-end and physical manufacturing. This is a vital "proof point" that Intel can, in fact, integrate its foundry services into the specific workflows of external companies.
Official Responses and Strategic Outlook
During the earnings call, CEO Lip-Bu Tan remained focused on the technical horizon, specifically the development of the 18A and 14A nodes. While analysts pressed for the names of major "anchor customers" to validate these nodes, Tan emphasized technical KPIs—defect density, transistor performance, and IP development—as the true markers of success.
CFO David Zinsner addressed concerns regarding the company’s increased capital expenditure, which is expected to exceed $20 billion in 2026. Zinsner framed the spending as a manifestation of confidence rather than a speculative gamble. "We’re going to be very careful around making bets ahead of customer commitments," Zinsner explained. "Given our confidence around next year, we must have pretty significant confidence in our customers, or we wouldn’t be putting purchase orders in place today."
However, this capital allocation remains complex because it serves both Intel’s internal products and its external foundry ambitions. With internal demand for Granite Rapids and Panther Lake chips reaching record levels, the fabs are currently well-utilized. The true test of the foundry model will come when Intel must balance these high-priority internal needs with the rigorous, contractual commitments of external clients during periods of industry-wide capacity constraints.
Implications: Optionality vs. Displacement
The broader industry implication is that Intel is currently positioning itself as a provider of "supply chain optionality" rather than an immediate challenger to TSMC’s market dominance.
Industry analyst Stephen Sopko of HyperFRAME Research argues that the market is not yet in a "winner-take-all" scenario. "TSMC is so overcommitted," Sopko noted. "Somebody else being able to produce chips and package them is a net positive. It’s not necessarily a competitive story between TSMC and Intel yet."
For many customers, the goal is diversification. By qualifying Intel as a second source or utilizing Intel’s advanced packaging services, companies can build resilience against geopolitical and supply chain shocks. Yet, this is merely an intermediate stage.
The Path to True Merchant Foundry Status
To reach the final goal, Intel must overcome two significant hurdles:
- The "Internal Rivalry" Perception: Customers must be convinced that their designs will not be deprioritized if Intel’s internal product groups face a supply crunch. This is the "trust deficit" that comes with being a vertically integrated company.
- Culture Shift: A successful merchant foundry must adapt its entire organization—from software tools and process design kits (PDKs) to customer support—around the customer. Historically, Intel’s fabs were designed for Intel’s own, highly predictable design cycles. Adapting to the chaotic, diverse requirements of the broader market remains an ongoing internal transformation.
As it stands, the second quarter of 2026 will be remembered as the moment Intel proved it could execute on its own manufacturing roadmap. The next chapters will be written by external customers. As Sopko aptly concluded, "What I need to see is somebody choosing Intel as the primary, with another fab providing optionality."
Until that happens, Intel remains a foundry in transition—highly capable, increasingly efficient, but still proving its place as a reliable, neutral partner in the global silicon ecosystem.
