Intel's contract chip-making business turned in a much-improved quarter, with $5.8 billion in revenue and a sharply narrowed loss. Yet only $293 million of that came from customers outside Intel — a stark reminder that the company's ambitious plan to become a merchant foundry is still more promise than proof.

For Windows users, PC builders, and IT pros, the gap between Intel's manufacturing progress and its commercial foundry success matters more than a balance sheet. It shapes which processors end up in your next laptop, how secure your company's network hardware will be, and whether the chip industry can finally break its dependency on a single geographic region.

The Quarter in Brief: Manufacturing Health, Customer Anemia

Intel Foundry's second-quarter 2026 numbers tell two stories. The factory operations are genuinely improving: revenue jumped from $4.4 billion a year ago, and the operating loss shrank from $3.2 billion to $2.1 billion. Margins, while still deeply negative, moved from -71.7% to -36.2%. Those gains reflect better yields, faster production cycles, and higher utilization across Intel 4, Intel 3, and the cutting-edge Intel 18A nodes.

But the external customer revenue — the money Intel earns from companies that aren't Intel — was just $293 million. That's up from a meager $174 million in the first quarter, but it's still a rounding error on a $5.8 billion topline. Intel remains overwhelmingly its own best customer, filling fabs with its own Core and Xeon processors.

This split matters because a true merchant foundry must attract outside chip designers — companies like Qualcomm, NVIDIA, or even Microsoft — to commit high-volume products to its factories. And that's not happening yet at anything close to a meaningful scale.

What's Actually Being Made on Intel 18A Right Now

Intel's most advanced node, 18A, entered high-volume production in January 2026 at Fab 52 in Arizona. It's a big deal because 18A introduces two critical innovations: RibbonFET gate-all-around transistors and PowerVia backside power delivery. The latter gives Intel a six-to-twelve-month lead over TSMC's equivalent technology, and it directly improves performance and efficiency.

But for now, the only chips rolling off 18A lines in volume are Intel's own. Two products anchor the node:

  • Panther Lake (Core Ultra 300 series): The next generation of laptop processors, packing up to 180 TOPS of AI performance and next-gen Arc graphics. This is the chip that will power premium Windows AI PCs when it launches.
  • Clearwater Forest (Xeon 6+): A server CPU with up to 288 efficiency cores, aimed directly at cloud data centers and enterprise racks running Windows Server workloads.

These products are why 18A exists today. They give Intel a reliable internal demand to fill the factory while the company irons out yield issues (currently in the 55–65% range, per reports) and works toward cost-level yields expected by late 2026.

For you, the Windows user or IT decision-maker, these chips matter. Panther Lake will dictate the performance and battery life of your next laptop; Clearwater Forest will drive the VMs and databases your organization depends on. A healthy 18A ramp means better availability and maybe more competitive pricing. A troubled one could lead to delays or supply constraints.

The One Real External Win: Fortinet's Security Processor

The most concrete evidence that outsiders trust Intel's fabs is a deal with cybersecurity giant Fortinet. The company plans to use Intel's proven Intel 4 process for its next-generation Security Processor 6 (SP6), which will power enterprise firewalls and network security appliances.

This isn't a cutting-edge gamble — Intel 4 is a mature node already used for Intel's own chips. But the engagement goes beyond a simple wafer order: Intel is providing back-end design services, packaging, and exclusive manufacturing. It's a hybrid foundry-plus-custom-silicon model that plays to Intel's strengths as a full-stack semiconductor company.

For IT teams managing Windows environments, the Fortinet win has a concrete payoff. A more efficient SP6 could lead to faster firewall throughput, deeper SSL inspection, and harder-to-exploit hardware roots of trust — all of which directly protect your Active Directory, hybrid cloud, and remote-access deployments. And if Intel establishes itself as a reliable supplier for such critical infrastructure, it could ease supply-chain fears that have long been concentrated around a single foundry.

How We Got Here: The Long (and Expensive) Road to IDM 2.0

Intel's foundry ambitions trace back to its “IDM 2.0” strategy launched in 2021. The goal was to open its fabs to the world while simultaneously catching up to TSMC in process technology. The company poured tens of billions into new factories in Arizona, Ohio, and Europe, betting that a combination of leading-edge nodes and geographic diversity would lure customers burned by TSMC's Taiwan-centric production.

The technical progress has been real. Intel 4 and Intel 3 brought it back to competitive parity, and 18A now offers a genuine performance-per-watt advantage — even if it trails TSMC's N2 in raw transistor density (about 238 million transistors per square millimeter vs. 313 million). But the commercial conversion has lagged. TSMC's ecosystem lock-in — its massive library of pre-validated design tools, IP blocks, and experienced engineers — means that switching foundries is a multi-year, multi-million-dollar commitment for any chip designer.

Moreover, Intel's own internal demand can be a double-edged sword. When Panther Lake and Clearwater Forest ramp up, they fill the fab. That's good for utilization but makes outside customers wonder: when capacity gets tight, will Intel prioritize its own chips over ours? The company has tried to address this by creating a separate foundry business unit with its own P&L, but trust is earned over several product cycles, not a single earnings report.

What You Should Do Now (Home Users, IT Pros, Enthusiasts)

If you're buying a Windows laptop in the next year: Panther Lake is coming, but don't wait indefinitely. Intel's current Meteor Lake and Lunar Lake chips are solid, and Qualcomm's Snapdragon X Elite offers strong ARM-based competition. If AI tasks (local Copilot, background blur, creative tools) matter to you, a Panther Lake device with its beefy NPU could be worth the wait — but expect prices to be high at launch.

If you manage enterprise or datacenter Windows deployments: Keep an eye on Clearwater Forest's rollout. The many-core architecture is designed for dense virtualization, a sweet spot for Windows Server. And if your organization uses Fortinet gear, ask your vendor about the SP6 roadmap; a hardware refresh timed with the new processor could boost security performance without a forklift upgrade.

If you're a PC builder or enthusiast: Intel's recent Arrow Lake desktop chips are already on Intel 3. The foundry's yield improvements mean more stable supply and possibly more competitive pricing against AMD. But don't expect foundry drama to directly affect your gaming rig — the real impact is on Intel's long-term ability to invest in R&D and fab capacity, which trickles down over years.

The Troubling 14A Gambit

Beyond 18A, Intel's future hangs on 14A, the next major node. CEO Lip-Bu Tan has made a momentous decision: Intel will only build out its massive Ohio fab for 14A if it can first secure major external customer commitments. That's a sharp break from the old “build it and they will come” approach. It's also a tacit admission that the merchant foundry thesis remains unproven.

14A is slated for risk production in 2028, but Intel has said it needs to see commitments by late 2026 or 2027. Reports suggest that Tesla and Apple are evaluating 14A for future AI and mobile chips, respectively, but nothing is signed. For the broader Windows ecosystem, a successful 14A could mean a healthy, competitive Intel cranking out next-next-gen CPUs for your PC. A shelved 14A would likely cede the leading-edge market entirely to TSMC and Samsung, with far-reaching consequences for chip pricing, innovation, and supply security.

Outlook: What to Watch Next

Intel's foundry story is at a pivot point. The execution is improving, but the strategy still needs large-scale external validation. Keep your eyes on these signposts:

  • Q3 and Q4 2026 earnings: Look for growth in external revenue beyond $300 million — and any named customer wins on 18A-P or 14A.
  • Microsoft Maia confirmation: If Microsoft commits its AI accelerator to Intel's fabs, it would be a massive vote of confidence. So far, it's only rumored.
  • Fortinet SP6 production ramp: When volume shipments begin, it will prove Intel can support a real commercial product.
  • 14A customer announcements: A disclosed commitment from a Tesla or Apple would transform the narrative overnight.

The next six months will tell us whether Intel Foundry is a niche experiment or the foundation of a more resilient semiconductor supply chain for the Windows world.