Qualcomm Just Opened the Door for Samsung at Its Most Important Foundry Battle

Qualcomm just handed Samsung an opening it has waited years for.

At the Snapdragon Summit 2026 in Maui last week, Qualcomm’s senior leadership drew a line under a relationship that has quietly been the anchor of the Android chip ecosystem for more than a decade — and then immediately pointed toward an ambiguity that could shift the balance of the global foundry market.

Chris Patrick, Qualcomm’s senior vice president and general manager of the Mobile Handset Business Group, did not mince words. “Samsung is the most powerful partner globally” for Qualcomm’s mobile business, he told reporters during a group interview. The qualifier matters: “most powerful” does not mean “only.” But in an era where Apple has moved almost all of its high-volume custom silicon away from Samsung, and TSMC dominates both the leading-edge logic node and the bulk of Qualcomm’s current production, the emphasis lands squarely on a familiar name.

Then came the second signal, and it was the one that will make people in semiconductor supply chains sit up.

The 2nm review that is not a review

Qualcomm confirmed it is conducting an “in-depth work” evaluation of Samsung Foundry’s 2nm process for future Snapdragon products.Patrick would not say whether the company had progressed to test chips or tape-out. He also would not rule it out. The language was careful but deliberate: Qualcomm is assessing how its next generation of chips could be produced at Samsung’s new node.

The current generation — Snapdragon 8 Elite Gen 6 and the Snapdragon 8 Elite Gen 6 Extreme — are both fabricated on TSMC’s 2nm process. That is no accident. TSMC has held Qualcomm’s flagship SoC production for years, and its process yield and performance edge have been a significant factor in why Samsung’s foundry push has stalled at the commercial level despite technical ambition.

But the fact that Qualcomm is now publicly evaluating Samsung’s 2nm — rather than quietly continuing its existing arrangement — changes the nature of the conversation. Supply chain diversification is no longer a vague hope; it is an active option on the table.

Why this matters beyond Korea

The implications extend well past Samsung’s Fab 1 in Hwaseong or its planned US fabrication complex in Texas. Two audiences will read this differently.

First, Apple. Qualcomm’s willingness to qualify Samsung at 2nm strengthens the case for the kind of dual-source foundry strategy that Apple already employs for memory and that some analysts believe it may consider for select logic products. Every additional fab that can deliver competitive 2nm performance raises the leverage of every chip design company over its fabrication partners.

Second, and more directly, the entire Android premium tier. Snapdragon is the reference chipset for most flagship Android devices outside China, and Samsung is both the largest Android vendor and the largest foundry competitor to TSMC in the world. If Qualcomm moves even a portion of its Snapdragon volume to Samsung 2nm, it validates the process for every other chip designer that has been watching Samsung’s advanced-node journey with cautious interest.

That validation effect is worth more than any single wafer lot.

The packaging shift nobody is talking about much

The summit also revealed a packaging change that deserves attention. The Snapdragon 8 Elite Gen 6 Extreme introduces an offset PoP — Package on Package — design. Instead of stacking the LPDDR6 memory directly above the SoC, the memory is shifted laterally. That opens space for a thermal interface material layer connecting the chip to the phone’s cooling system, lowering thermal resistance and allowing sustained high clock speeds for longer periods.

The practical consequence is that Qualcomm is treating heat — not just transistor density — as the binding constraint on flagship performance. This matters for the broader industry because it signals a shift in how mobile SoC design philosophy is evolving. As AI workloads move onto device and token generation rates become the real bottleneck, packaging architecture is becoming as important as process node.

Qualcomm also acknowledged that it is actively exploring PIM (Processing-in-Memory) and HBC (High-Bandwidth Compute) approaches to address the same memory-bandwidth problem from different angles. Neither is near commercialization, but both represent strategic bets on the next decade of chip architecture — and both require close partnership with foundry and memory producers. Samsung, which manufactures both foundry-capable chips and LPDDR memory, occupies a structurally advantageous position in that equation.

The LPDDR6 cooperation that ties two giants together

Another detail in the interview is easy to overlook but strategically significant. Qualcomm and Samsung Memory jointly developed LPDDR6 — with both companies participating in theJEDEC standardization process and engineering teams collaborating on new signaling methods and memory-channel management.

The benchmark reference phone at the summit used SK Hynix LPDDR5X, and Qualcomm declined to confirm whether it had validated SK Hynix’s LPDDR6. That non-answer itself is telling: Qualcomm is keeping its memory supply options open while deepening its primary partnership with Samsung.

This bilateral memory collaboration is not trivial. LPDDR6 will be the standard for the next generation of AI phones, and whoever sets the timing and volume for that transition will define the pricing and availability landscape for everyone else. Samsung’s dual role as both a foundry and a memory producer gives it a rare positioning that no other foundry — especially not TSMC, which does not produce memory — can replicate.

What happens next

Three things are likely in the near term.

First, Samsung Foundry will use this endorsement aggressively. The company has struggled to win major design wins at its most advanced nodes since TSMC pulled ahead several generations ago. Qualcomm’s public framing of Samsung as its “strongest partner” — coupled with a confirmed 2nm evaluation — will be cited in investor presentations, customer meetings, and recruitment campaigns for months. The message is simple: the top Android chipmaker sees a future at Samsung.

Second, TSMC will feel immediate competitive pressure. Not existential pressure — Qualcomm has not announced any volume move yet — but enough to trigger defensive engagement. Expect TSMC to intensify its own customer outreach, possibly accelerating process improvements or offering more aggressive terms for next-generation Snapdragon volumes. The foundry business runs on momentum, and any perception of erosion invites motion from competitors.

Third, and perhaps most importantly, the dual-source narrative gains traction across the industry. Every chip designer that has been burned by single-fab dependency — and there are plenty — will take note. The trend toward multi-source fabrication, once a niche risk-mitigation strategy, is becoming a mainstream requirement. This is good for designers. It is less good for fabs that cannot compete on price, performance, or delivery speed simultaneously.

The super-cycle question nobody can answer yet

Qualcomm’s leadership also touched on what it hopes will be the demand-side driver of all this: the AI agent phone and the replacement cycle it could stimulate. Patrick declined to predict when a “super cycle” might arrive, comparing the exercise to forecasting stock markets. But his underlying thesis was clear — as on-device AI agents learn to use phones more fully, the value proposition of a high-spec device increases, and consumers replace sooner.

If that thesis proves correct, the demand for capable flagship chips rises, and the question of who makes them becomes financially material for every player in this story. Qualcomm, Samsung, TSMC, Apple, and the Chinese OEMs all have stakes in the outcome.

What is clear today is that Qualcomm has reopened a door that appeared closed. Samsung is walking through it. The rest of the industry is watching to see how wide it gets.