Intel Foundry has finally put a corporate customer's name on a press release under CEO Lip-Bu Tan. Cybersecurity giant Fortinet signed on to manufacture its next-generation Security Processor 6 on the Intel 4 process node, pulling a piece of custom silicon production away from Taiwan Semiconductor Manufacturing Company. On paper, poaching a long-standing TSMC client sounds like the long-awaited turning point for a foundry business that lost billions while chasing external volume. In reality, reading past the headline reveals a far more modest trade-off that highlights how much ground Santa Clara still needs to cover.
The core issue facing Intel's foundry ambition has never been a lack of ambition or promises; it has been the total absence of commercial clients willing to risk their primary product lines on Intel's manufacturing floor. Fortinet's decision to entrust its SP6 ASIC to Intel 4 gives the foundry team a valid victory to show Wall Street. Yet, because Intel 4 is a legacy node that Intel itself has largely moved past for its premier processors, the agreement functions more as a low-risk stepping stone for both parties than a ringing endorsement of Intel's leading-edge capabilities.
A Poached Client with a Conservative Playbook
Fortinet commands roughly half of the world's firewall appliance shipments, making its custom Application-Specific Integrated Circuit strategy one of the most successful silicon plays outside the standard GPU market. For two decades, the security firm built its moat on proprietary chips rather than relying solely on off-the-shelf processors from traditional vendors. When Fortinet built its previous SP5 processor, it chose TSMC's 7-nanometer process. Moving the SP6 to Intel represents a clear loss for TSMC, but it is not a loss that threatens Taiwan's manufacturing supremacy.
For Fortinet, switching to an American foundry provides geographic diversification and supply chain resilience. That matters to enterprise customers and government agencies that demand secure hardware sourcing.
Fortinet is not gambling its business on an unproven line.
By opting for Intel 4, Fortinet selects an established 7nm-class node. It is a mature manufacturing process with well-understood yields and fully debugged toolsets. That makes total sense for a security chip, where raw node shrinkage takes a backseat to predictable thermal profiles, high yields, and stable long-term availability.
The Economic Math Behind the Announcement
To evaluate what this agreement actually means for Intel's balance sheet, consider how chip revenues flow through firewall hardware.
Assume a hypothetical cybersecurity hardware vendor generates $2 billion in annual appliance sales. In enterprise networking, hardware costs typically account for a fraction of final revenue, with software licenses, maintenance, and support contracts taking the lion's share. The actual silicon cost within each appliance might represent 5 percent to 10 percent of the total hardware bill of materials. Therefore, even a dominant market share in firewall chips yields tens of millions of dollars in foundry revenue per year, not the billions required to offset the staggering cost of modern extreme ultraviolet lithography plants.
Intel needs anchor clients capable of ordering hundreds of thousands of wafers on bleeding-edge nodes every quarter. Fortinet brings strategic credibility, but it does not supply the sheer volume required to fill empty fabs.
Why the Flagship Nodes Remain Vacant
The public dynamic inside Intel's foundry unit has changed under Lip-Bu Tan. Previous executive communications often focused on grand target dates and multi-node roadmaps. Tan took a quieter, more pragmatic stance, acknowledging that corporate clients were testing Intel's lines while refusing to disclose company names until formal production commitments were locked in place.
The central problem remains unaddressed by this announcement. Intel's regulatory disclosures have repeatedly confirmed a glaring absence of major external commitments for its 18A and 14A nodes—the exact technologies intended to dethrone TSMC at the top of the performance stack.
Hypothetically, if a major smartphone or AI accelerator designer shifted its flagship line to Intel 18A, it would require absolute certainty regarding yield rates, intellectual property protection, and physical delivery schedules. A failure on a flagship product line can cost a tech company billions in lost revenue within a single quarter. Major chip designers remain cautious because Intel is both a foundry vendor and a direct market competitor in server and PC CPUs.
Building trust across the semiconductor sector takes years of consistent execution. Winning a security ASIC contract on an older node proves that Intel can package, fab, and deliver custom hardware for an external partner. It demonstrates that the business mechanics of contract manufacturing are functional inside Santa Clara.
Yet, until a tier-one hyperscaler or fabless titan trusts Intel with its primary revenue driver on a leading-edge node, the foundry business remains an expensive experiment supported by internal production and government subsidies. The Fortinet agreement is a step forward, but the real test for Lip-Bu Tan's tenure has barely begun.
