Intel Foundry Lands Fortinet. But the Silence Around the Process Node Is the Real Headline

On a gray Tuesday morning in Lisbon, I was halfway through my second espresso when the news alert hit. Intel Foundry had named its first publicly confirmed external customer. Not Apple. Not Nvidia. Not Microsoft. It was Fortinet—the cybersecurity hardware company that most consumers have never heard of but every office network trusts. The coffee went bitter in my mouth. Not because the news was bad, but because it was too carefully staged. I have spent the better part of a decade decoding chip announcements, and this one smelled like a diplomatic photo op. Two executives smiling. A vague promise of “advanced technology.” A heavy emphasis on “security” and “trusted supply chain.” But no process node. No capacity volume. No timeline. In the foundry world, that silence is not a missing detail. It is the story. So let me tell you what Fortinet actually means for Intel—not as a transaction, but as a signal in the new geopolitical semiconductor cold war. To understand this deal, you need to go back to the moment when Intel lost its manufacturing swagger. For nearly five decades, Intel was the American chip champion. It designed and fabricated its own processors. Then TSMC pulled ahead in process technology, and Intel, stumbling through a series of delayed node names, became the industry’s favorite cautionary tale. The company eventually announced a radical plan: reinvent itself as a foundry, a contract manufacturer for the entire world. This was never a modest ambition. It meant competing with TSMC, which controls roughly sixty percent of the global foundry market, and Samsung, which controls another ten. It meant convincing rivals like AMD, Nvidia, or Apple to trust a company that had spent decades as their enemy. It meant pouring billions into new fabs in Arizona, Ohio, and Ireland. It meant accepting that the foundry segment would bleed money for years—and perhaps forever if the customers did not show up. The new CEO inherited that expensive promise. Every investor call, every strategy memo, every press event carried the same implicit question: “Who will actually use Intel Foundry?” For a long time, the answer was “internal products only.” Then Fortinet appeared. Fortinet is a fabless semiconductor company, but not the kind that makes headlines. It designs FortiASIC processors for its own security appliances—firewalls, unified threat management devices, edge routers. These chips do packet inspection, encryption, load balancing. They are not glamorous. They do not train large language models. They sit inside metal boxes in corporate server rooms, humming quietly, blocking attacks. In a world obsessed with AI accelerators, Fortinet feels almost nostalgic. But that ordinariness is exactly why the partnership is revealing. Here is what we actually know from the announcement. Intel Foundry and Fortinet have signed a multi-year agreement. Intel will manufacture Fortinet’s security chips in the United States. That is about it. No node. No volume. No financial terms. No date for first production. To anyone who has read a supply chain contract, that silence is the most informative part of the document. Start with the process node. Fortinet’s security processors are not performance monsters. A firewall chip needs high throughput and low power, but it does not need the absolute bleeding edge of N2 or 18A. Most of the workload is parallel packet processing, not single-thread speed. In my experience auditing chip roadmaps, a mature node like Intel 16 or Intel 3 would be more than sufficient. The marketing language in the announcement, heavy on “trust” and “resilience,” deliberately avoided the word “leading-edge.” That tells me the node is probably not 18A. And if it were 18A, Intel would have shouted it from the top of every fab in Phoenix. The fact that they did not is a strong statistical indicator of a mature process. This matters because the entire narrative of Intel Foundry’s pitch to the market rests on 18A being the great equalizer—the GAA-based, RibbonFET-powered node that would match or beat TSMC’s N2. The Fortinet deal does not validate that claim. It validates something else: that Intel can make reliable, secure, non-cutting-edge chips for an American customer. Nice, but not revolutionary. Yield is the next quiet worry. Intel has been burned by public reports of low yields on advanced nodes. The 18A node is still not fully proven at scale. Fortinet, as a mid-sized fabless player, does not have the engineering muscle to debug yield issues on a brand-new node. If Intel pushed 18A on them, the risk would be mutual. Fortinet could miss its product cycle, and Intel would face another black eye. A mature node is the safe choice. The absence of any yield commentary in the announcement is another hint that the node is boring enough not to need discussion. Packaging is the silent detail that usually tells you how strategic a foundry deal really is. Intel likes to talk about advanced packaging—EMIB, Foveros, 2.5D stacking—as its differentiator. The Fortinet announcement includes no mention of packaging. If the deal involved chiplets or multi-die security processors, Intel would have said so. Instead, silence again. This is probably a simple ASIC on a traditional package. That reduces the technical uniqueness of the deal. Capacity is the elephant in the room. Intel is spending like a daredevil who just inherited a rocket factory. The company has multiple fabs under construction across the United States, in Arizona, Ohio, and New Mexico. Capital expenditure has run at thirty to fifty percent of revenue for years, and dividends have been slashed to protect that spending. The depreciation load from these new fabs is enormous, and the foundry segment has been generating deep operating losses. A single customer like Fortinet—even if it fills a small percentage of one fab—will not move the profit needle. The financial effect on Intel is negligible in the short term. The strategic effect, however, is larger than any dollar figure. Supply chain dynamics are where the deal starts to get interesting. Fortinet’s biggest customers include governments and large enterprises. In the post-Snowden, post-cyberwar era, “made in the United States” is no longer just a marketing slogan; it is a procurement requirement. Many federal agencies now mandate that critical security components come from trusted suppliers in allied countries. This is not a theoretical preference. It is written into contracts. For Fortinet, using Intel Foundry turns its entire product line into a “secure American stack” story. It can tell its government buyers: we design in San Jose, and our chips are made in Arizona, not in Taiwan or China. That is worth real money in an RFQ, even if it does not improve firewall latency by a single nanosecond. Market demand is steady, not explosive. Cybersecurity hardware grows at a predictable rate, perhaps eight to twelve percent per year, driven by attack surface expansion and regulatory compliance. There is no AI-accelerator explosion here. But there is a structural shift toward nationalizing the semiconductor supply chain. As Japan, Europe, and the United States pour subsidies into local fabs, a small but stable demand wedge is opening for “trusted foundries.” Intel Foundry is positioned to capture that wedge. Fortinet is the first named brick in that wall. The competitive picture is still brutal. TSMC owns the foundry oligopoly. Samsung is second. Intel’s external foundry market share is somewhere around one to two percent, which is embarrassing for a company of its size. But the race is not solely about market share. The race is about credibility. Every external customer that commits to Intel—even a niche security ASIC customer—chips away at the assumption that “nobody serious uses Intel Foundry.” That credibility effect compounds. When procurement teams see Fortinet in the customer list, a CFO in a due diligence call might think: “If a cybersecurity company can entrust its chips to Intel, maybe we can too.” The financial reality is unglamorous but important. Intel’s foundry business has not disclosed granular gross margin for this deal, but the segment’s overall margins are abysmal. If 18A yields are not world-class, each wafer costs too much, and the foundry loses money. A mature node with reasonable volume could actually be more profitable than a flagship node. In that sense, the Fortinet deal might be financially smarter than chasing Nvidia with 18A. But it will still be small. The real financial answer will not arrive for another two or three years, when production reaches meaningful volume and Intel reports the foundry segment’s results. The geopolitical undercurrent is the part that gets lost in benchmark talk. The United States is no longer optimizing for the cheapest or even the fastest silicon. It is optimizing for survivability. The CHIPS Act was not designed to save Intel; it was designed to save America from a supply chain chokehold. Every dollar of subsidy is a bet that the US must be able to build the chips its defense and critical infrastructure depend on. Fortinet is a defense-adjacent company. Its appliances guard power grids, government networks, banks, and airports. For those customers, the origin of the silicon is not a technical specification—it is a threat model. Intel’s fabs in Arizona and Ohio sit on home soil, under US jurisdiction. That is a geopolitical insurance policy no Taiwanese foundry can fully match, even if TSMC builds a fab in Phoenix, because the technology and the most advanced R&D still live near the Taiwan Strait. The narrative repair signal is real, too. Do not underestimate how much of this deal is about restoring employee and investor morale. Intel has laid off thousands of people over the past two years. It cut dividends. It watched its market cap get dwarfed by Nvidia. The foundry division has been a money pit. The Fortinet announcement is a carefully placed chess piece in the new CEO’s campaign to prove that Intel Foundry can attract outside customers. It gives the workforce a headline to repeat: “First major customer.” It gives the board a talking point. It gives analysts a reason to ask for more details. In a crisis of confidence, that is worth more than the actual wafer revenue. Now for the contrarian angle. Almost everyone will frame this as “Intel wins a customer.” But the more I analyze the signal, the more I think Fortinet is using Intel as a hedge—not a technology destination. Fortinet’s existing supply chain is likely concentrated in TSMC, like almost every fabless company. The Intel agreement lets Fortinet build a second source for its most government-sensitive chips, ensuring that if geopolitical chaos hits the Taiwan Strait, Fortinet can still ship firewalls to the Pentagon. That is not a vote of confidence in Intel’s process technology. It is a vote of confidence in Intel’s geography. This is the fork in the road where code met chaos and won. The intersection was not a specific node or a benchmark score. It was the moment when the US government decided that semiconductor supply chains are a matter of national security. Once that decision was made, the value of “domestic production” began to exceed the value of “best-in-class performance.” TSMC will build a fab in Arizona, yes. But TSMC’s headquarters, its most advanced R&D, and its centralized supply chain remain in Taiwan. For a customer as exposed to Washington politics as Fortinet, that is a risk no PowerPoint can paper over. The blind spot in the coverage is that Intel is not really competing with TSMC on a level playing field. It is competing with the Taiwan problem. And the Taiwan problem is a permanent feature of the global map. Intel can say, “Everything from lithography to final test is on American soil, under American jurisdiction, with no foreign factory in between.” That pitch is worth more to Fortinet than a ten percent improvement in performance per watt. The second blind spot is the hole in the press release. If this were 18A, Intel would have named the node. The absence of that name should temper all the excitement. A mature-node, low-volume arrangement proves nothing about Intel’s ability to compete on the leading edge. It becomes an elaborate public relations exercise. Investors who hope this is the beginning of a wave that will eventually include Nvidia or Apple will be disappointed. Intel needs one big, advanced-node customer—not a steady but modest security vendor—to truly enter the foundry big league. Fortinet is not that customer. But that is okay, at least for now. Intel does not need to win the entire industry in one day. It needs to win a corner of the industry where trust matters more than transistors. Fortinet is that corner. It is the first brick in a wall that could protect Intel from irrelevance. So where do we go next? I will be watching three markers over the coming quarters. The next Intel Foundry Direct Connect event—if 18A is ever mentioned in the same breath as Fortinet, the order may be larger than currently expected. Fortinet’s quarterly earnings calls—any mention of “Intel” or “US-made silicon” in the prepared remarks will tell us whether this is a strategic supply chain move or a footnote. And the emergence of a second customer with a similar profile—another American security or defense-adjacent chip designer—would confirm the thesis that Intel is building a trusted-foundry ecosystem, not a TSMC killer. I believe the most likely path is the quiet one. Intel Foundry will slowly collect a portfolio of trustworthy but unglamorous customers. Fortinet will be the first. Aerospace, defense, networking, and government-oriented chip companies will follow. The world will not see a giant line of ARM servers being manufactured by Intel, at least not in the next three years. But the world will see the word “resilience” become the new “performance” in semiconductor marketing. The fork in the road where code met chaos and won is now visible in every foundry contract that crosses a border. The real winner of the Fortinet deal is not Intel’s product roadmap. It is the idea that, in a world where chips are weapons, the country that controls the fab controls the firewall. Fortinet just placed its firewall, and its supply chain, on American soil. I will keep my coffee hot for the next announcement.