Arm is canceling Qualcomm's chip design license

Original Article ↗ Hacker News Discussion ↗

Nature of the ARM–Qualcomm dispute

  • Dispute centers on Qualcomm’s use of custom ARM cores derived from its Nuvia acquisition under an architectural license, vs ARM’s standard “buy our cores” model.
  • ARM claims Nuvia’s favorable, non‑transferable license was server‑only and can’t simply move to Qualcomm; cancelled Nuvia’s license and plans to cancel Qualcomm’s architectural license after trial.
  • Qualcomm argues its broader license should cover the work and that ARM is overreaching to protect high-margin core-licensing revenue.
  • Exact contract terms and how much Oryon reuses Nuvia IP are unclear from the thread.

Qualcomm’s strategic options

  • Short term: still allowed to license ARM reference cores (Cortex/Neoverse), so phones and many products can keep shipping “as usual.”
  • Medium term: likely outcomes seen as (a) settlement with higher royalties; or (b) gradual shift toward RISC‑V while keeping ARM cores as a bridge.
  • Some argue Qualcomm has more to lose (Android OEMs could switch to MediaTek/Samsung); others say ARM is more exposed because Qualcomm is probably one of its largest revenue sources.

RISC‑V as alternative

  • Many see this fight as a major tailwind for RISC‑V and note Qualcomm is already investing heavily in it and pushing extensions to ease ARM→RISC‑V reuse.
  • Others say high‑end RISC‑V is years behind: ecosystem, toolchains, SIMD/vector maturity, and flagship‑class SoCs are still emerging.
  • Debate whether swapping an ARM front‑end for RISC‑V on an existing microarchitecture is “relatively small” work or a deep redesign touching MMU, memory model, CSRs, verification, etc.

Performance and technical debates

  • Current RISC‑V hardware is widely viewed as competitive for microcontrollers and embedded, but far behind Apple M‑series and top ARMv9 mobile/server cores.
  • Counterpoint: ISA isn’t the bottleneck; high‑end out‑of‑order RISC‑V cores exist as IP (e.g., from several vendors) but haven’t yet appeared in mass‑market products.
  • Long technical subthreads discuss compressed instructions, decode complexity vs x86, vector ISA design (RVV vs alternatives), and front‑end width.

Ecosystem and OS considerations

  • Android: already has some RISC‑V work; most apps are bytecode, but a large share of top apps ship native code via the NDK, so widespread recompilation or emulation would be required.
  • Windows: ARM support took over a decade to become usable; commenters expect Windows on RISC‑V to be even slower to mature despite Microsoft’s early interest.
  • x86 emulation is seen as essential for any non‑x86 Windows platform, but there’s tension between relying on emulation vs driving native ports.

Impact on mobile SoC market

  • Android OEMs are not “stuck”: can source from MediaTek and Samsung, or keep buying ARM‑core‑based Qualcomm chips. MediaTek Dimensity and recent Exynos parts are seen as viable, though often still behind top Snapdragon/Apple in performance/efficiency and modem quality.
  • Some argue ARM’s move risks making it look hostile and proprietary, pushing large customers faster toward RISC‑V; others say ARM must enforce contracts and defend its ecosystem from one vendor becoming too dominant.

Perceptions of ARM’s role

  • ARM is portrayed both as:
    • Protecting its ecosystem and preventing fragmentation by one dominant custom‑core vendor; and
    • An IP landlord trying to preserve high‑margin core licensing in the face of architectural‑license customers and a rising open ISA competitor.
  • Several note SoftBank’s ownership and IPO pressures likely drive ARM’s more aggressive posture.