护城河对比
最多三家已覆盖的公司，逐项对照。每一项判断都是人工编辑的判断，而非公司披露的数字——并且每一项都附有其引用依据。
| Marvell Technology | Micron | AMD | |
|---|---|---|---|
| 护城河评级 | 窄 FY2026 10-K (filed 2026-03-11): differentiated platform IP — over 10,000 issued patents and pending applications as of 2026-01-31, plus a proven custom ASIC platform leveraging ultra-high-speed SerDes, silicon photonics, co-packaged optics and custom HBM — but Marvell itself calls its markets 'intensely competitive' with 'pricing pressures', notes customers 'have chosen to develop certain semiconductor products internally', and discloses two >=10% customers with the ten largest at 82% of FY2026 net revenue. Real, defensible IP in a concentrated, contestable customer base = narrow, not wide. https://www.sec.gov/Archives/edgar/data/1835632/000183563226000011/mrvl-20260131.htm | 窄 Micron sits in a three-firm DRAM oligopoly (with Samsung and SK hynix), holds "over 60,000 patents" and leading-edge process technology (industry-first 1-gamma EUV DRAM node in FY2025), yet its FY2025 10-K concedes competitors "may use aggressive pricing" and "may have a larger market share," and consolidated gross margin swung from negative 9% (FY2023) to 40% (FY2025) — a real but commodity-cyclical, non-durable advantage. | 窄 AMD's FY2025 10-K documents real but bounded advantages: approximately 7,200 U.S. patents and roughly 18,900 patent matters worldwide, an outright claim that 'We are the market share leader in semi-custom game console products,' and gross margin of 50% on revenue up 34% to $34.6 billion. The same filing sets the ceiling: 'Some of our competitors may possess stronger market positions, larger customer bases, more design wins, and greater financial, sales, marketing, and distribution resources than us'; Nvidia is named 'the discrete GPU market share leader'; and AMD relies on TSMC 'for the production of all wafers for microprocessor and GPU products at 7 nanometer (nm) or smaller nodes.' Advantages that are genuine, contested, and dependent on a shared foundry are narrow, not wide. |
| 护城河类型 | 无形资产与知识产权 The moat rests on hard-to-replicate mixed-signal IP: the 10-K describes the custom ASIC platform built on ultra-high-speed SerDes, ARM compute, security, storage, silicon photonics and advanced packaging (die-to-die interconnects, chiplets, CPO, custom HBM), with multiple 5nm designs executed, 3nm in progress and a 2nm platform in development; a secondary switching-cost element comes from multi-year custom design wins co-developed to individual customer specifications. https://www.sec.gov/Archives/edgar/data/1835632/000183563226000011/mrvl-20260131.htm | 无形资产与知识产权 The 10-K roots Micron's edge in proprietary technology and IP: "over 60,000 patents" granted as of Aug 28, 2025, "significant investments to develop proprietary product and process technology," the industry's first 1-gamma (EUV) DRAM node, ninth-generation G9 NAND, and HBM4 samples to key AI customers. | 无形资产与知识产权 The FY2025 10-K locates the durable asset in design IP rather than manufacturing: 'We rely on contracts and intellectual property rights to protect our products and technologies from unauthorized third-party copying and use,' with approximately 7,200 U.S. patents and about 18,900 patent matters worldwide spanning x86 EPYC/Ryzen CPUs, CDNA/RDNA graphics and the Versal/Zynq adaptive-SoC families, and it credits customer wins to 'our broad IP portfolio and leadership in design, integration and advanced packaging.' It is explicitly not cost_scale: AMD is fabless and 'utilize[s] Taiwan Semiconductor Manufacturing Company Limited (TSMC) for the production of wafers' — the same leading-edge foundry its merchant competitors use, so no manufacturing-scale advantage accrues to it. |
| 领先地位 | 并列领先者 Leader in its optics niche, #2 in custom silicon: ~60% of high-end PAM4 DSP share (36kr, 2026-06-27, https://eu.36kr.com/en/p/3870758441178373) but an estimated 20-25% of custom AI ASIC design services versus Broadcom's ~70% (hashrateindex, 2026-05-13, https://hashrateindex.com/blog/design-partners-ai-asic-market-part-2/) — net, a co-leader in AI data-center connectivity/custom silicon behind Broadcom overall. | 并列领先者 A member of the three-supplier DRAM oligopoly and at the technology frontier (industry-first 1-gamma EUV node), but the smallest of the three by share — the 10-K acknowledges competitors "may have a larger market share" — and a later entrant scaling up in HBM behind SK hynix and Samsung. | 快速追随者 The FY2025 10-K itself places AMD behind the pace-setter in its two most important markets: 'Our principal competitor in the supply of discrete graphics is Nvidia, who is the discrete GPU market share leader,' and the risk factors state Nvidia 'leverages its market position in data center GPU, financial resources, and proprietary software ecosystem to promote its systems and influences customers who do business with us.' AMD is nonetheless executing 'an annual cadence of leadership for AMD Instinct solutions' and has won gigawatt-scale commitments (OpenAI, 6 GW) — following fast and closely, not setting the pace. |
| 定价权 | 中等 The 10-K characterizes Marvell's markets as having 'pricing pressures' and intensifying competition (https://www.sec.gov/Archives/edgar/data/1835632/000183563226000011/mrvl-20260131.htm), yet the Q1 FY2027 release reports 52.1% GAAP / 58.9% non-GAAP gross margin on record revenue (https://www.sec.gov/Archives/edgar/data/1835632/000183563226000014/q127_8kx522026ex-991.htm) — differentiated-IP margins, tempered by hyperscaler buyer power. | 弱 Memory is a commodity with cyclical ASPs: the 10-K flags "volatility in average selling prices" and competitors' "aggressive pricing" as key risks, and consolidated gross margin swung from -9% (FY2023) to 22% (FY2024) to 40% (FY2025), marking Micron largely as a cyclical price-taker; contract-based, supply-tight HBM is a partial exception. | 中等 FY2025 gross margin was 50%, up from 49% in 2024 per the 10-K — healthy and improving — but the same filing attributes average-selling-price pressure to competitors: Intel 'uses its microprocessor market position to price its products aggressively and target our customers and channel partners with special incentives. These aggressive activities have reduced and may reduce our unit sales and average selling prices for many of our products'. Pricing power is real but contested. |
| 综述 | Marvell is a fabless data-infrastructure silicon supplier whose center of gravity has shifted decisively to the AI data center: the data center end market was $6,100.3M, 74% of FY2026 revenue, up from 40% two fiscal years earlier (FY2026 10-K). Its strongest position is electro-optics — in high-end PAM4 optical DSPs for 400G+ transceivers it holds roughly 60% share on Inphi-inherited SerDes/FEC IP, with Broadcom above 30%, the two together over 90% (36kr, 2026-06-27). In custom AI silicon it is the structural #2 design partner at an estimated 20-25% of the custom AI ASIC design-services market versus Broadcom's ~70%, anchored by AWS Trainium and Microsoft Maia wins (hashrateindex, 2026-05-13). The Q1 FY2027 release (2026-05-27) shows the flywheel turning — record $2.418B revenue (+28% YoY), Q2 guided to $2.7B mid-point (+35% YoY), management citing 'exceptional AI-related bookings' across 800G/1.6T optics, 51.2T switches, CPO/NPO and custom XPU — and the Celestial AI (Photonic Fabric) and XConn (PCIe/CXL switching) acquisitions closed in February 2026 extend the interconnect moat toward scale-up fabrics. The offsets that keep the moat narrow are in Marvell's own filing: intense competition (AMD, Alchip, Astera, Ayar, Broadcom, Credo, GUC, Lightmatter and others), hyperscaler in-housing risk, and heavy customer concentration. | Micron is one of only three scaled DRAM suppliers (with Samsung and SK hynix) and a leading-edge NAND maker, defended by proprietary product and process technology and more than 60,000 granted patents. In FY2025 it shipped the industry's first 1-gamma (EUV) DRAM production node and sampled HBM4 36GB 12-high to multiple key customers for next-generation AI platforms, evidencing process-technology leadership in AI memory (FY2025 10-K). | AMD's edge is architectural design IP executed on someone else's fabs. The FY2025 10-K describes a full-stack portfolio — EPYC server CPUs, Instinct AI accelerators, Ryzen client parts where 'AMD was the first company to integrate a dedicated neural processing unit (NPU) on the same SoC as an x86 CPU for AI PCs,' Radeon graphics, Pensando networking and Versal adaptive SoCs — backed by roughly 18,900 patent matters worldwide, and it discloses gigawatt-scale customer commitments (an October 2025 agreement with OpenAI 'to deploy 6 gigawatts of AMD GPUs,' first gigawatt on Instinct MI450). The same filing bounds that moat: Nvidia is 'the discrete GPU market share leader' and 'leverages its market position in data center GPU, financial resources, and proprietary software ecosystem'; Intel 'uses its microprocessor market position to price its products aggressively'; Arm architectures and customers who 'internally develop products to support similar AI workloads' are named as encroaching; and every wafer at 7 nm or below comes from TSMC. Revenue grew 34% in FY2025 with Data Center up 32%, but the filing's own competitive framing places AMD in the challenger seat rather than the entrenched one. |
| 产业链位置 | Fabless supplier spanning 'data center core to network edge': it sits between hyperscaler AI compute (custom XPU/XPU-attach ASICs) and the optical layer (PAM4/coherent DSPs, CPO/LPO, DCI, AEC, PCIe retimers), outsourcing fabrication to independent CMOS foundries; the Feb-2026 Celestial AI and XConn acquisitions push it further into scale-up photonic fabric and PCIe/CXL/UALink switching (FY2026 10-K, https://www.sec.gov/Archives/edgar/data/1835632/000183563226000011/mrvl-20260131.htm). | US-based supplier of the memory tier of the AI stack — DRAM, HBM and NAND; the 10-K notes cloud servers for AI and data-centric workloads "require significantly increasing quantities of DRAM, including HBM," and Micron sampled HBM4 to power next-generation AI platforms. | A fabless merchant-silicon designer sitting one layer above the foundries: the FY2025 10-K states AMD relies on TSMC 'for the production of all wafers for microprocessor and GPU products at 7 nanometer (nm) or smaller nodes' and primarily on GLOBALFOUNDRIES above 7 nm, with packaging and test performed by Asia-Pacific ATMP partners — so its cost, supply and cadence are inherited from partners it does not control. |
| 产品（份额 / 壁垒） |
|
|
|
| 长周期投票 | +0.13 权重 0.20 · 群体判断 bullish 编辑先验，未经回测。 | +0.05 权重 0.20 · 群体判断 bullish 编辑先验，未经回测。 | +0.06 权重 0.20 · 群体判断 bullish 编辑先验，未经回测。 |