Compare moats

One-of-a-kind silicon sold to very few buyers: the IPO prospectus discloses G42 at 24.0% of 2025 revenue (85.0% in 2024), MBZUAI at 62.0% of 2025 revenue, and a December 2025 MRA with OpenAI representing 'a substantial portion of our projected revenues over the next several years' — against a competitive field the same document names as NVIDIA, AMD, Intel, hyperscaler accelerators and the major clouds.

source: sec.gov

The FY2026 10-K (filed 2026-08-27) shows IREN holds an input the industry competes on: it names 'access to secured and energized power' first among what it believes are the principal competitive factors in its industry, and reports 'executed grid connection agreements, letters of agreement or equivalents representing approximately 5GW of total power capacity' in the United States, Canada, Spain and Australia as of June 30, 2026. It does not claim an advantage over rivals in that input. It says 'Certain competitors may have access to more competitively priced power, a greater access to power and a better capacity to timely secure grid connections,' and 'Many of our competitors have greater financial, technical or commercial resources, longer cloud operating histories or larger customer bases, compared to us.' Operating AI Cloud Services capacity was approximately 40MW at June 30, 2026, and the stored XBRL fundamentals from this 10-K show an FY2026 operating loss of about $1,046.7M on revenue of about $707.0M, so no durable advantage is yet demonstrated.

source: sec.gov

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

source: sec.gov

Wafer-scale integration is the moat's substance: the prospectus contrasts the WSE-3's 4 trillion transistors on 46,225 mm² of silicon with NVIDIA's B200 at 208 billion on roughly 1,600 mm² — a commercial chip category with one occupant.

source: sec.gov

The FY2026 10-K's asserted advantage is control from owning its data centers, 'including the associated land, grid connections and substations', which it believes allows it 'to benefit from more sustainable cash flows and operational flexibility relative to operators that rely upon third-party colocation services or short-term land leases'. Its only scale language is a belief that its 'procurement scale, deployment experience and direct control over the data center layer' lets it bring new compute into service 'rapidly and at scale', a deployment-speed claim, and it 'generally target[s]' regions with 'low-cost and attractive renewable energy sources', a siting target rather than a demonstrated cost position. It does not claim a cost or scale lead: it says 'Many of our competitors are larger, have longer operating histories and significantly greater resources than we do' and that certain competitors 'may have access to more competitively priced power, a greater access to power and a better capacity to timely secure grid connections.' There is no network effect or lock-in either, since existing contracts have 'terms ranging from month to month up to five years'. No enumerated moat source is grounded.

source: sec.gov

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

source: sec.gov

Scoped to the category the record supports: wafer-scale integration has no other commercial occupant per the prospectus's own comparison, and the Q2 release claims — attributed, via AMD and AWS partnerships — leadership in disaggregated inference, a technique the prospectus defines.

source: sec.gov

The 10-K lists IREN's competitors as hyperscalers (Amazon Web Services, Google Cloud, Microsoft Azure, Oracle Cloud) and specialized AI Cloud Services providers (CoreWeave, Nebius, Crusoe, Lambda, Nscale, SpaceX and others). It says IREN began providing AI Cloud Services in 2024 and cites its 'shorter operating history in AI Cloud Services relative to some competitors', and states 'Some of our competitors have longer operating histories, larger customer bases, more comprehensive IP portfolios and patent protections, more design wins, and greater financial, sales, marketing and distribution resources than we do.' Operating AI Cloud Services capacity was approximately 40MW at June 30, 2026. The filing describes a smaller, later entrant that trails some competitors on operating history, customer base and resources, with its progress so far resting on two anchor contracts (a ~$9.7B five-year Microsoft agreement and a ~$3.4B five-year NVIDIA contract) rather than any claimed category position.

source: sec.gov

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.

source: sec.gov

The CFO states Q2 'significantly improved core gross and operating margins compared to a year ago' — improving, from a base where three named customers hold most of the revenue and the leverage that comes with it.

source: sec.gov

The 10-K's risk factors state 'Our competitors' products, services and technologies may be cheaper or provide better functionality or features than ours, which has resulted and may in the future result in lower-than-expected selling prices or demand for our products.' They also say long-term contract pricing 'is generally fixed or agreed at the time of contracting', so if market pricing for comparable capacity rises during a contract's term 'we will not benefit from those increases with respect to capacity already committed under our existing long-term contracts', and that if customers suffer a downturn or discontinue its services it 'may be compelled to offer more flexible terms, lower our prices or risk losing a significant customer.' The stored XBRL fundamentals from this 10-K show FY2026 revenue of about $707.0M against an operating loss of about $1,046.7M, so there is no margin record yet that evidences pricing power.

source: sec.gov

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.

source: sec.gov

Layer-4 wafer-scale accelerators plus an inference cloud — hardware vendor and specialized cloud in one node.

Downstream AI-cloud operator: owns grid-connected data centers and deploys NVIDIA/AMD GPU systems it rents to hyperscalers, frontier labs, AI developers and enterprises (anchor customers Microsoft and NVIDIA).

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).

• CS wafer-scale systems (WSE-3) Leader · Deep source: sec.gov
• Cerebras inference cloud Challenger · Moderate source: sec.gov

• AI Cloud Services (bare metal and managed GPU compute) Unknown · Moderate source: sec.gov
• AI infrastructure software (Mirantis k0rdent AI) Unknown · Low source: sec.gov
• Bitcoin mining (in wind-down) Unknown · Low source: sec.gov
• Owned data centers and grid-connected power portfolio Unknown · Moderate source: sec.gov

• Custom AI XPUs & ASICs Top 3 · Deep source: hashrateindex.com
• Data center interconnect & CPO/LPO Unknown · Deep source: sec.gov
• Ethernet switching (scale-out & scale-up) Unknown · Moderate source: sec.gov
• PAM4 & coherent optical DSPs Leader · Deep source: eu.36kr.com
• Storage controllers & fibre channel Unknown · Moderate source: sec.gov

Editorial prior, not backtested.

see exactly how it voted →

Editorial prior, not backtested.

see exactly how it voted →

Editorial prior, not backtested.

see exactly how it voted →