Skip to content
TheChatArchiveThe Chat Archive
o3-2025-04-16· OpenAIHugging Face LMArena Human Preference 140K

What is the story of the c1 modem and why did it take 6 years?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-7-sonnet-20250219’s answer.
  1. User

    What is the story of the c1 modem and why did it take 6 years?

  2. Assistant

    Below is the “short version” first.
    Keep scrolling if you’d like the full-blown timeline and the technical deep-dive into why building a first-class cellular baseband takes so long— even for Apple.

    ─────────────────────────────────────────────── TL;DR
    • “C1” is Apple’s code-name for its first custom 5G/LTE modem.
    • Work began in late-2018/early-2019 (right after Apple’s legal war with Qualcomm and the purchase of Intel’s failing modem unit).
    • Original launch target: 2023 iPhone 15.
    • Reality: slipped to 2026 at the earliest.
    • Six years (and counting) because making a modern, carrier-certified, multi-band, multi-RAT (2G → 5G mmWave) modem is insanely hard: RF + DSP + firmware + worldwide certification + patent minefield.
    • Apple is still using Qualcomm modems under an extended supply deal while C1 matures.

    ───────────────────────────────────────────────

    1. Context — why Apple wanted its own modem
      • Cost & margins: Qualcomm collects ~$7–9 in royalties per iPhone plus the silicon cost.
      • Control: total vertical integration is Apple’s playbook (A-series CPU/GPU, M-series Macs, custom ISP, etc.).
      • Legal bad blood: the 2017-19 “Qualcomm v. Apple” lawsuit over FRAND licensing.
      • Risk: Intel’s 5G project was crumbling; Apple needed a long-term hedge.

    2. Key milestones (and missed ones)

    YearWhat happened“How it went”
    2018Apple secretly spins up “Project Siberia” (internal 4G/5G modem effort).Skeleton crew only.
    Apr 2019Apple & Qualcomm settle lawsuits; 6-year license + chipset supply signed.Buys Apple time.
    Jul 2019Apple buys Intel’s smartphone-modem business for $1 B.2,200 engineers + patents.
    2020Work on the first silicon (C1, on TSMC N6) starts.Covid slows on-site RF lab work.
    2021Early C1 samples hit Apple’s internal boards.Failing carrier conformance tests (LTE fallback, power draw).
    2022Rumors of 2023 iPhone launch.Modem heat + mmWave throughput issues; timeline slips.
    Sep 2023Apple quietly re-ups with Qualcomm through 2026.Public signal that C1 is not ready.
    2024Second-gen spin (“C2”) taped-out on TSMC N4P.Lab results better; still not carrier-grade.
    2025/26*Earliest realistic window (iPad or Apple Watch first, then iPhone).*Industry analysts’ best guess.
    1. Why it really takes 6+ years

    A. Engineering complexity
    • A CPU/GPU SoC is mostly digital logic. A modem is a hybrid monster:
    – RF/analog front-end (from 600 MHz low bands to 28 GHz mmWave).
    – Mixed-signal baseband + ultra-hard-real-time DSP pipelines.
    – Tens of millions of lines of firmware.
    • Has to support every legacy tech: 2G (still used for emergency in some countries) → 3G → LTE → 5G NR sub-6 → 5G mmWave, plus VoLTE, VoNR, dual-SIM/DSDS, emergency call stacks, eSIM, GNSS assistance, carrier aggregation, etc.

    B. World-wide compliance gantlet
    • 3GPP conformance → PTCRB → GCF → regional regulators (FCC, CE, MIC, NCC, …) → per-carrier field certification (Verizon’s LBS, AT&T’s NumberSync, China Mobile’s IMS, etc.).
    • Hundreds of thousands of test vectors covering RF emissions, SAR, call-drop, hand-over, VoLTE latency, etc. One failure means more silicon spins.

    C. Patent & licensing minefield
    • Qualcomm, Nokia, Ericsson, InterDigital, Huawei each hold portfolios critical to 5G.
    • Apple must design-around to shrink royalty stack, while staying standards-compliant, and still negotiate cross-licenses. Legal reviews alone add months.

    D. Power and thermal budget
    • The iPhone PCB is a packed SiP. A modem draws >2-3 W during 5G uplink bursts; even a ~10 % inefficiency shows up as heat and battery drain.
    • Apple’s first C1 reportedly consumed ~30-40 % more power than Qualcomm’s X65 under identical conditions—deal-breaker for an iPhone launch.

    E. RF front-end supply chain
    • Power-amplifier modules, filters, switches, antenna tuning — mostly 3rd-party (Qorvo, Skyworks, Murata). Apple has to refactor those around its own baseband pin-outs and calibrations.

    F. Org & process learning curve
    • Apple’s silicon culture is “CPU-centric”. A carrier modem team needs RF PhDs, systems engineers, field-test groups, standards negotiators. Intel’s group came with baggage: legacy code, morale issues, disparate tool chains. Integration and re-tooling take years.

    1. Where things stand now (mid-2024)
      • Apple still uses Qualcomm’s X70 (iPhone 15) and is expected to use X75 (iPhone 16).
      • Internal chatter says the first commercial outing of C-series might be a lower-risk product (cellular iPad or an Apple Watch Ultra with satellite fallback) before it hits the flagship iPhone.
      • Apple will keep iterating until power, performance, and certification metrics match or beat Qualcomm. The moment they cross that threshold, expect an iPhone launch followed by legal renegotiations of royalties.

    2. Bottom line
      Building a best-in-class cellular modem is among the most specialized and regulation-heavy feats in consumer electronics. Even a $3-trillion company with a billion-dollar head start and 2,000 veteran engineers needs ~6–8 years to (hopefully) ship a competitive chip. The “C1 story” is Apple learning—painfully—why Qualcomm’s moat is so wide and why almost every other smartphone OEM still buys modems rather than rolling their own.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.