Qualcomm’s CEO Cristiano Amon handed Intel a blunt reality check on September 5 when he told Bloomberg that Intel “is not an option today” for making Qualcomm’s Snapdragon X chips, while adding that “we would like Intel to be an option.”
The short, pointed assessment lands at a pivotal moment for Intel’s long-run bet to become a major contract foundry, and it exposes a widening gap between Intel’s manufacturing ambitions and the supply relationships of the biggest chip designers.
Amon’s comment matters because Qualcomm’s Snapdragon X family is not a fringe product. Those chips are manufactured on TSMC processes tuned for mobile and mobile-class laptops; Qualcomm’s designs, paired with efficient nodes, are already powering Arm-based notebooks that compete on power efficiency with the best x86 silicon. That makes Qualcomm not only a potential customer Intel badly wants, but also a visible barometer of whether Intel’s process technology measures up in real product terms.
Intel’s roadmap is under heavy scrutiny. The company has placed a huge strategic wager on its 18A node and an ultimate return to leadership, while warning it may halt or delay 14A development if it cannot attract major external customers or hit internal milestones. Recent reporting and industry commentary point to yield and execution problems on 18A and other advanced steps, raising the stakes for Intel’s pitch to outside designers. Amon’s public dismissal narrows Intel’s near-term options for landing marquee foundry customers.
Why the rest of the industry’s biggest design houses keep circling TSMC
Apple, Nvidia, and AMD have kept the world’s most advanced nodes tightly booked with TSMC for years, and those relationships show why Intel faces an uphill climb.
• Apple has repeatedly leaned on TSMC for its flagship A- and M-series chips, using TSMC’s most advanced process variants as a strategic lever to extract performance and power gains for iPhones and Macs. That partnership has helped cement TSMC as the go-to supplier for premium mobile and client SoCs.
• Nvidia’s highest-end GPUs — including Hopper-family dies and the H100 — have been manufactured on customized TSMC processes (TSMC 4N variants), and Nvidia continues to coordinate closely with TSMC on cutting-edge packaging and node work to scale transistor counts and performance for AI workloads. That deep technical coupling is hard to replicate quickly.
• AMD has similarly relied on TSMC for multiple generations of CPUs and GPUs, using the foundry’s 5nm/4nm family to keep parity or lead with performance and energy efficiency. Those supply commitments make it difficult for a new entrant to pry away high-volume design partners.
Taken together, these partnerships show a pattern: top-tier fabless and integrated designers choose foundries that already offer proven yields, node maturity, and packaging ecosystems. For Intel to break into that circle, it must do more than announce roadmaps; it has to demonstrate reliable yields, attractive cost per transistor, and production scale that matters to customers used to TSMC’s timetable.
The irony in Intel’s story is visible in product roadmaps. Intel is planning client-class chips (Nova Lake and related lines) that will, in some cases, mix Intel’s own nodes with TSMC’s N2—an arrangement that underlines both the technical interdependence that still exists and Intel’s current inability to fully replace TSMC for its highest-performance tiers. That reality makes Intel’s pitch to firms like Qualcomm harder to swallow: why move away from a supplier already delivering at scale?
What Qualcomm’s stance signals
Amon’s “not an option today” line is a tangible vote of no confidence for Intel’s near-term foundry story. It also sends a message to the market: Qualcomm will stay where the product meets its performance and efficiency targets. The door is left open for Intel if and when Intel can prove it has the process maturity and yields Qualcomm needs, but for now, Qualcomm is staying with proven suppliers.
For Intel, turning statements and roadmap slides into repeatable production at scale is an execution problem that demands capital, time, and — crucially — anchor customers willing to take early risk. Apple, Nvidia, and AMD are not that kind of early-stage customer anymore; they deploy the world’s leading nodes and partner tightly with the foundry that has delivered them for years.
Against this backdrop, public rejections from heads of major chip designers are expected to keep the pressure on the Santa Clara firm until it can show sustained results on nodes customers care about.
