event-trade登录编辑
瑞银Codex · 实盘不做把握 5/10
北京时间·🕒 发布 08-03 17:43:23时区推断·⟳ 抓取 08-03 18:05:26·⚡ 决策 08-04 14:09:12 +20h03m·84.1 天前

瑞银称 Arm 授权收入和数据中心版税改善,但手机版税走弱且下调目标价

这是一篇卖方财报后更新,核心事实来自已披露的 Arm 季报和指引,新增交易信息有限。

📄 原文瑞银↗ 原文
Arm Holdings PLC: Licensing Growth Offsets Handset Royalties Weakness
Jun Q again beat consensus on stronger-than-expected licensing revenue (ex-SoftBank), while Sep Q revenue guidance came…
不做 Arm Holdings(ARM.US)
不做:卖方财报后更新方向混合,业绩与指引事实已公开交易,目标价反而下调,缺少可量化的新订单或未定价催化。
不做 AMD(AMD.US)
不做:仅为 Arm 报告中的泛 AI CPU 需求读穿,没有 AMD 自身订单、客户定点或收入增量。
不做 英特尔(INTC.US)
不做:读穿逻辑过泛,无法证明短期收入、利润或订单的公司级增量。
不做 博通(AVGO.US)
不做:AI networking 受益表述属于行业景气评论,未给出新订单或相对市值足够的量化增量。
不做 Marvell Technology(MRVL.US)
不做:没有 Marvell 自身新增事实,单篇卖方读穿不足以形成可回测交易。
不做 Astera Labs(ALAB.US)
不做:受益链条来自泛 AI 网络主题,未坐实公司级订单、客户或业绩弹性。
展开详情(成交明细 / 研究档案 / 原文全文 / 决策记录链)

研究档案

🔎 新鲜度:最早出现 2026-08-03 · 扩散 medium · price-in 风险 high ↗

关键理由

  • 这是一篇卖方财报后更新,核心事实来自已披露的 Arm 季报和指引,新增交易信息有限。
  • 报告同时上调 FY27/FY28 收入与 EPS 估计、强调 AI 授权和数据中心版税强劲,但又下调 CY28 EPS 和目标价,方向不够单边。
  • 对 AMD、英特尔、博通、Marvell、Astera Labs 的读穿只是泛 AI CPU/网络景气评论,未给出订单、客户定点或可量化新增收入。

主要风险

  • 若后续 Arm 或客户侧披露明确 AGI CPU 订单、晶圆供给或客户名单,可能形成新的独立催化。
  • 单篇卖方报告不适合作为短线开仓依据,尤其财报数据已被市场交易近一个完整交易日。

原文全文

展开原文 · 5,432 字
Arm Holdings PLC: Licensing Growth Offsets Handset Royalties Weakness
Jun Q again beat consensus on stronger-than-expected licensing revenue (ex-SoftBank), while Sep Q revenue guidance came in ~3% above expectations at the midpoint. On the royalty side, Jun Q DC (CPU + networking) royalties more than doubled Y/Y again (a trend ARM expects to continue over the next few Qs). Meanwhile, handset units and mix are sounding incrementally weaker as ARM suggested that the softness is now spreading from low-end devices into mid-range and premium Android handsets that carry higher royalty rates. Apple's unit resilience is not that helpful, given lower royalty rates vs Android. This leads us to lower our FY27 royalty growth expectation to 17% Y/Y from ~20-21%. Licensing comes to the rescue here, as ARM argues that current strength is being driven by AI-related chip development rather than pull-ins. Licensing revenues are set to grow ~30% Y/Y in FY27E - notably above the low-double-digit CAGR expectation we have heard in the past. On AGI, demand is now exceeding the $2B level discussed 90 days ago, though supply remains the constraint. We are raising our FY28 AGI revenue estimate to $1.2B from $1.0B but modestly lowering GM assumptions given likely higher wafer costs. Net-net, we are raising FY27/FY28 revenue and EPS estimates modestly, though our CY28 EPS estimate moves slightly lower on lower gross margins, resulting in a PT reduction from $360 to $320.
Analysis/Key Highlights
Results: ARM reported Jun Q revenue of $1.29B in the upper half of the guidance range modestly above UBSe/Street and EPS of $0.45, $0.01 above the high end of the range and $0.05 above UBSe/Street. Revenue grew 22% Y/Y as royalties of $715MM came in largely in-line with our estimates and 2% above the street and licensing of $574MM was 2%/3% above UBSe/Street. Opex was $733MM - below UBSe and Street, driving operating margin of 41.2% (+200bps Y/Y and -800bps Q/Q).
Guidance
ARM guided Sep Q revenue of $1.38B +/-50MM, up ~7% Q/Q at the midpoint and 3% above the Street, opex of ~$780MM, and EPS of $0.43-0.51 vs. prior UBSe/Street of $0.45/$0.44. Management expects both royalty revenues to grow ~13% Y/Y (or down -2% Q/Q) and license revenue to grow roughly +30% Y/Y or +17% Q/Q.
Handset IP
ARM called out incremental slowdown in the smartphone royalties weighing on the FY27 royalty expectations. Different from its expectations, ARM is now seeing weakness broadening beyond the low-end segment and impacting mid-range and high-end categories as well. Our own work suggested Apple handset units remained most resilient, but also carry lower royalty rates vs Android ecosystem players, particularly CSS customers (we think Samsung, MediaTek). As a result, ARM now expects FY27 royalty revenues to grow high-teens vs prior expectation of ~20%.
Data Center IP
DC IP royalties doubled Y/Y and ARM expects those to continue doubling over the next few Qs and act as a big part of the offset for the handset royalty weakness. Cloud AI represented 14% of FY26 royalty revenues or ~$365MM and should it more than double in FY27, it could be on track for $750MM+ in FY27 (or ~24% of the royalty revenues). We expect that NVDA and AWS will maintain more favorable rates vs other CPU IP customers like MSFT, GOOG, but understand all royalties are charged per core with royalty rates trending higher gen-to-gen. ARM also highlighted SuperNICs and DPUs adding to the royalty revenues.
Licensing
Similarly to Mar Q, licensing revenues came in above expectation, but importantly Sep Q outlook, at ~90MM above Street/UBSe, now suggests licensing revs will grow ~100MM Q/Q while ARM continues to expect a 40/60 1H/2H revenue split. This implies a near-30% Y/Y increase - ahead of royalties and notably above the longer-term CAGR of LDD at best that ARM discussed in the past. ARM is seeing strong momentum for its CSS programs and, to a lesser extent, conversion of particular customers to total access / ATA packages.
AGI CPU
ARM reiterated AGI revs of ~$100MM in F4Q27, but suggested that it is working hard to increase FY28 AGI revenues above the $1B target announced at the ARM Everywhere event as it now sees demand in excess of $2B, while supply remains the key bottleneck. We understand that AGI CPU will be offered in 2 main SKUs: a custom device for META and a merchant device for other customers like OAI, OCI, CBRS and others. We think GMs may vary and likely slightly lower for the custom program. ARM suggested that among options of securing additional wafers might be repurchasing wafers from other players who have TSMC allocation, which might have a higher cost base.
FY2027 Outlook
We continue to expect ARM's revenues to grow >20% Y/Y in FY2027 at ~25% (UBSe) as we expect licensing revenues to grow ~30% Y/Y and royalties ~17% Y/Y. We model GM of 97% for FY27E, impacted by the CPU revenue contribution in 4Q. We model FY27E OpEx of $3.2B up 18% Y/Y.
Read-Across
Bullish commentary on CPU demand and custom silicon complexity, benefiting AMD and INTC, as well as AVGO, MRVL, and ALAB which benefit from broader AI networking opportunities.
Valuation: PT $320 (from $360)
We derive our $320 price target based on a PEG methodology. We use a 1.8x target PEG (unchanged). Our LT EPS CAGR for Arm is 42%; applying our 1.8x PEG multiple results in a 78x P/E multiple. 78x our 2028E EPS of $4.14 (vs prior $4.33) gives a $320 price target.

决策记录链 · 只追加

每次领取产生一条记录,重新决策追加一条并以 supersedes 指向被替代的记录(ADR 0014)。

这条决策早于决策账本(2026-10-08 起逐条追加),没有追加记录。