Compare moats
Up to three covered companies, band by band. Every call is a curated editorial judgment, never a disclosed figure — and every band carries its cited basis.
| Iron Mountain | Equinix | GitLab | |
|---|---|---|---|
| Moat rating | narrow Iron Mountain's physical-records franchise shows a real switching-cost advantage, but the FY2025 Form 10-K also describes a slowly shrinking need for it, and the growth businesses compete in open markets. On the advantage side, an independent record: the Department of Justice complaint published in the Federal Register on 2016-04-11 (https://www.federalregister.gov/documents/2016/04/11/2016-08210/united-states-v-iron-mountain-inc-and-recall-holdings-ltd-proposed-final-judgment-and-competitive) stated that "Iron Mountain is the largest RMS company in the United States" and that "Taken together, permanent withdrawal fees and other withdrawal restrictions make it difficult for a new RMS entrant to win customers away from existing RMS vendors." The 2025 10-K reports more than 740 million cubic feet stored, says "we have consistently experienced strong customer retention levels", and shows the Global RIM segment's Adjusted EBITDA margin at 44.7% in 2025 against 44.6% in 2024. On the limiting side, Item 1A says "Our Records Management and Data Management service revenue growth is being negatively impacted by declining activity rates as stored records and tapes are becoming less active and more archival" and "A significant shift by our customers to storage of data through non-paper or non-tape-based technologies, whether now existing or developed in the future, could adversely affect our businesses." In data centres the 10-K says it competes with "numerous data center developers, owners and operators, many of whom own properties comparable to ours". A strong, regulator-documented lock-in on a physical market the company itself expects only to hold steady, plus growth arms without that lock-in, is a narrow moat rather than a wide one. | wide The FY2025 10-K (filed 2026-02-11) grounds the advantage in an asset that took 27 years to assemble and that a competitor cannot buy: "Over our 27-year history, we have curated a diverse, industry-leading ecosystem of more than 500,000 interconnections", "over 10,500 customers, including 2,000+ network service providers and a leading market share of cloud-on ramps", across "280 data centers, in 77 markets in 36 countries", with "99.9999%+ operational uptime" delivered in 2025 and no single customer at 10% of revenue. Because the value of each IBX rises with who else is already inside it, incumbency compounds rather than decays. The counterweight is real and disclosed: Item 1A says "The global multi-tenant data center market is highly fragmented. It is estimated that we are one of more than 2,400 companies that provide these offerings around the world", and warns that competitors "may adopt aggressive pricing policies". That caps the pricing that the moat converts into, not its durability — the fragmentation sits in commodity space-and-power, while the interconnection density the filing describes has no comparable substitute. FY2025 revenue of $9.217B with operating margin recovering to 20.0% from 15.2% in FY2024 is consistent with the incumbency holding. | narrow The FY2026 10-K (year ended 2026-01-31) shows real expansion inside the installed base: 118% Dollar-Based Net Retention, Base Customers up to 10,682 from 9,893, customers above $100,000 of ARR up to 1,456 from 1,229, and $1.0 million ARR customers "to 155 from 123, an increase of 26%". But the filing concedes "limited barriers to entry" and names "Microsoft Corporation, which owns GitHub" as principal competitor. Retention decelerates (130% FY2024, 123% FY2025, 118% FY2026), and since DBNR is reported "on a threshold basis of 130%," the FY2024 figure is a cap and the slide is steeper than it looks. Narrow, not wide: expansion must be re-won against a bundled hyperscaler. |
| Moat type | switching costs The advantage is the cost and friction of moving stored records. The 10-K's Item 1A says that "if we are successful in winning record storage customers from competitors, the process of moving their stored records into our facilities is often costly and time consuming", and its notes record payments made "to a customer in order to terminate the customer's storage of records with its current records management vendor" as customer inducements. The DOJ complaint (https://www.federalregister.gov/documents/2016/04/11/2016-08210/united-states-v-iron-mountain-inc-and-recall-holdings-ltd-proposed-final-judgment-and-competitive) describes the mechanism: "Customer contracts also often impose a cap on the number of boxes per month that a customer may permanently remove from a RMS vendor's facility, such that a switch to a new RMS vendor may take several months to complete", and a new vendor "will have to offer to pay the fees to induce the customer to switch". Scale helps, since the 10-K says "the majority of our competitors operate in only one market or region", but the filing's own account of what keeps customers is retention and the cost of leaving, which makes switching costs the primary source. | network effects The 10-K names the mechanism itself, twice and unprompted: "As more customers choose Equinix for high connectivity and performance reliability at the metro edge, it benefits their suppliers and business partners to colocate in the same data centers and connect directly with each other. This adjacency creates a network effect that attracts new customers while continuously enhancing our value proposition to existing customers", and in the Competitive Landscape section, "This ecosystem creates a network effect that improves performance and lowers the cost for our customers". The evidence is the count of participants rather than any patent or unit-cost claim — 500,000+ interconnections, 2,000+ network service providers, a leading share of cloud on-ramps, an Internet Exchange the filing calls "the largest global peering solution". Switching costs are a genuine second layer (fixed-duration contracts billed on space and power, physical cross connects into resident counterparties), but they are what holds a customer already inside the ecosystem; the reason to enter in the first place is who is already there. | switching costs Stickiness comes from platform embedding, not IP. The 10-K rests differentiation on "our single platform with a unified data model," on "Consolidation of multiple tools into a single platform," and on deployment inside the customer's own perimeter for regulated buyers. Replacement means re-integrating a toolchain and re-certifying compliance. IP is the wrong label: only "16 issued patents and 25 pending patent applications," plus open-source licenses granting "broad permissions to use, copy, modify, and redistribute." The no-lock-in claim is the filing's own characterisation, footnoted to a Forrester study "commissioned by GitLab." |
| Leadership | clear leader The band applies to records management, the Global RIM segment that produced $5,291,481 thousand of 2025's $6,901,737 thousand revenue. Independent evidence: the DOJ complaint published in the Federal Register on 2016-04-11 (https://www.federalregister.gov/documents/2016/04/11/2016-08210/united-states-v-iron-mountain-inc-and-recall-holdings-ltd-proposed-final-judgment-and-competitive) stated "Iron Mountain is the largest RMS company in the United States", with fiscal 2014 worldwide revenues of "approximately $3.1 billion", and called Recall, with 2014 revenues of "approximately $836.1 million", "the second-largest RMS company in the United States"; the FY2025 10-K lists property "acquired in connection with our acquisition of Recall Holdings Limited". The 2025 10-K adds, in the company's own words, that "the majority of our competitors operate in only one market or region". The independent source is from 2016 and covers the United States only. Iron Mountain did go on to acquire Recall, but the proposed final judgment in the same notice required it to divest "Recall RMS assets in all fifteen geographic markets identified in the Complaint", thirteen of them to Access CIG, which the notice called "currently the third-largest RMS provider in the United States", so Recall's business was not absorbed whole. This leadership does not extend to data centres, where the Global Data Center segment's $803,429 thousand of 2025 revenue compares with total revenue of $9,217,000,000 at Equinix (https://www.sec.gov/Archives/edgar/data/1101239/000110123926000032/0001101239-26-000032-index.htm) and $6,112,692,000 at Digital Realty (https://www.sec.gov/Archives/edgar/data/1297996/000110465926015365/0001104659-26-015365-index.htm) in stored fundamentals from their FY2025 10-Ks. | clear leader Leadership is claimed on interconnection, not on square footage, and the filing's own evidence is about density: "our position is unmatched in the industry" is supported by 2,000+ resident network service providers, "a leading market share of cloud-on ramps", 500,000+ interconnections and an Internet Exchange described as "the largest global peering solution", across 36 countries. The band is read against neutral, ecosystem-dense colocation, where that footprint has no direct analogue. It is deliberately not read against total data centre capacity: the same 10-K puts Equinix among "more than 2,400 companies" in a "highly fragmented" MTDC market, and Item 1A concedes the company must compete for land and power against "new market entrants" drawn in by AI. | fast follower The 10-K positions GitLab against a leader rather than as one: "Our principal competitor is Microsoft Corporation, which owns GitHub." Competition reads from the challenger side, differentiating "from GitHub through flexible deployment options that work within enterprise security and compliance requirements, LLM neutrality with self-hosted gateway support" and an open core model. On AI it follows: Duo Agent Platform went "Generally available in January 2026," the last month of the fiscal year. What keeps it a follower rather than behind is cadence, "a new version of our software every month for 172 months in a row," plus a regulated-deployment franchise a SaaS-first rival does not cover. |
| Pricing power | strong In records, growth comes from price on steady volume. MD&A attributes Global RIM "organic storage rental revenue growth driven by revenue management" (organic storage rental growth of 5.4% in 2025), while Item 1A says volumes "were relatively steady in 2025" and "Our organic revenue growth has been positively impacted by our ability to effectively introduce, expand and monitor revenue management. If we are not able to continue and effectively manage pricing, our results of operations could be adversely affected". The segment's Adjusted EBITDA margin was 44.7% in 2025 against 44.6% in 2024. In data centres MD&A cites "improved pricing" and "a 620 basis point increase in Adjusted EBITDA Margin reflecting recent lease commencements, improved pricing and cost containment", to 51.8% from 45.6%. The limits: Item 1A says that on power costs "we may be limited in our ability to, or may not always choose to, pass these increased costs on to our customers", and the 2016 DOJ complaint (https://www.federalregister.gov/documents/2016/04/11/2016-08210/united-states-v-iron-mountain-inc-and-recall-holdings-ltd-proposed-final-judgment-and-competitive) found that where Iron Mountain and Recall met, they "competed aggressively against one another for customers, resulting in lower prices for RMS". Adjusted EBITDA margin is a non-GAAP measure, and the Q2 2026 release (https://s204.q4cdn.com/148941814/files/doc_financials/2026/q2/FINAL-Q2-2026-Earnings-Press-Release.pdf) shows it at 35.8% for the company against 36.7% a year earlier. | moderate Contract structure supports price: fixed-duration agreements billed on space and power plus per-connection interconnection fees, an installed base too physically entangled to move cheaply, and 99.9999%+ uptime in 2025 as the thing being paid for. FY2025 revenue of $9.217B grew about 5% on FY2024's $8.748B while operating margin recovered to 20.0% from 15.2%, so pricing and cost were at least held. But the filing itself refuses the strong band: competitors "may adopt aggressive pricing policies, especially if they are not highly leveraged or have lower return thresholds than we do. As a result, we may suffer from pricing pressure that would adversely affect our ability to generate revenues", and some rivals bundle communications or cloud services against bare colocation. Power procurement is a further pass-through risk the filing flags. Price is defended, not dictated. | moderate FY2026 gross margin was 87% against 89%, with cost of revenue up $35.6 million "primarily due to an increase of $18.4 million in third party hosting costs for SaaS and cloud usage," and management expects SaaS and Duo mix to bring costs that "may adversely impact our gross margins." Strain is explicit: GitLab "may be required to reduce our prices," competitors "may offer their products and services at a lower price or for free," and it "implemented user limits on our free SaaS product." Power retained: Premium lists at $29 per user/month billed annually, and seats bundle $12 and $24 of GitLab Credits. But "A majority of our subscriptions are on a one-year period." |
| Summary | Iron Mountain stores and manages physical records for, per its FY2025 10-K, more than 240,000 customers in 61 countries, "including approximately 95% of the Fortune 1000", with "no single customer accounting for more than approximately 3% of revenue". Revenue was $6,901,737 thousand in 2025, of which the Global RIM segment contributed $5,291,481 thousand and the Global Data Center segment $803,429 thousand. The records business is the moat. Storage rental agreements "generally range from one to five years in length", the company says it has "consistently experienced strong customer retention levels", and a 2016 Department of Justice complaint described Iron Mountain as "the largest RMS company in the United States" and permanent withdrawal fees as a barrier to rivals winning customers. That lock-in shows up as pricing: MD&A attributes Global RIM storage growth to "revenue management" while Item 1A says volumes "were relatively steady in 2025". The threat is substitution: activity is declining as records become "less active and more archival", and a shift to non-paper storage "could adversely affect our businesses". The growth arms sit outside that moat. The company operated 31 data centers across 21 markets with 488 MW of capacity approximately 97% leased and a weighted average lease expiration of 10.3 years, counts "five of the largest global hyperscalers among our customers", and on 2026-08-05 reported data center leasing of 110 megawatts year to date and second-quarter revenue of $2.0 billion, up 18.5%. But in data centres it competes with operators owning "properties comparable to ours", and in asset lifecycle management with hyperscalers' own programmes. A durable switching-cost franchise in a market expected only to hold steady, funding growth businesses that compete on power, location and price, is a narrow moat. | Equinix is a network-neutral, multi-tenant colocation and interconnection REIT: it does not sell compute, it sells the metro-edge real estate where networks, clouds and enterprises physically meet. The FY2025 10-K describes the platform as "280 data centers, in 77 markets in 36 countries" serving "over 10,500 customers, including 2,000+ network service providers and a leading market share of cloud-on ramps", carrying "more than 500,000 interconnections" curated over 27 years, with 61% of 2025 revenue recognised outside the U.S. Revenue is structurally recurring — infrastructure offerings are "billed based on the space and power a customer consumes" under fixed-duration contracts generating MRR, interconnection is "billed based on the outbound connections from a customer" — and no customer reached 10% of 2025 revenue. AI enters the story as demand rather than as a product: the filing positions Equinix as the interconnect point for "model providers, data platforms, neoclouds and gateways", and pushes core hyperscale capacity into xScale, which is "developed and operated through our joint venture partnership arrangements". The bear case is in the company's own Item 1A. The MTDC market is "highly fragmented", Equinix being "one of more than 2,400 companies"; competitors "may adopt aggressive pricing policies"; the AI build-out invites "significant investments in the data center industry by both current competitors and new investors", after which "we could lose market share" and must "compete against certain of these competitors to secure the land and power needed for our expansion plans". Product extension has also failed before — the filing notes past offerings "have been or are being discontinued, including the Equinix Metal product". The honest reading: the interconnection ecosystem is close to unreplicable and the moat sits there; the capacity business around it is a capital race Equinix enters with scale but no immunity. | GitLab sells a single-platform DevSecOps toolchain defended by workflow lock-in rather than intellectual property: a unified data model plus deployment flexibility, including single-tenant "Dedicated for Government with FedRAMP compliance," makes it hard to remove where data residency is contractual. The installed base expands: 118% Dollar-Based Net Retention, 10,682 Base Customers, 1,456 above $100,000 of ARR, 155 above $1.0 million (up 26%), over 70% of ARR from public sector and enterprise. The filing supplies the counterweight: "limited barriers to entry," Microsoft/GitHub as principal competitor, 16 issued patents. Retention (130 to 123 to 118, the 130 a cap), gross margin (89% to 87%) and a January 2026 AI launch point to compression at the commodity end. No source measures market share, so every share band is "unknown". |
| Chain position | In the AI build-out Iron Mountain is a landlord and a recycler. It leases data center capacity to hyperscale customers, with "five of the largest global hyperscalers among our customers", and its ALM business "provides hyperscale and corporate IT infrastructure managers with services and solutions that enable the decommissioning, data erasure, processing and disposition, and recycling or sale of IT hardware and component assets". Upstream, Item 1A says "We rely on third parties to provide power to our data centers" and that construction depends on "one or more design firms, general contractors, and associated subcontractors". | Equinix is the neutral meeting point of the AI and cloud supply chain rather than a link in its manufacturing path: it houses other companies' compute and sells the adjacency between them. The 10-K places it between the network layer (2,000+ service providers), the cloud layer ("a leading market share of cloud-on ramps") and enterprise consumers who "assemble these capabilities into operational stacks", and describes an AI ecosystem "of model providers, data platforms, neoclouds and gateways" curated for enterprise AI demand. Core hyperscale training capacity sits beside that, not inside it, in xScale, built with JV partners so hyperscalers "add to their core hyperscale data center deployments and existing customer access points at Equinix". The revenue is therefore levered to AI's distribution and inference edge more than to training-cluster buildout. | GitLab sits at the software-tooling layer, not compute or models. It consumes third-party LLM capacity — marketing "LLM neutrality and support for self-hosted AI gateways, including air-gapped environments" — so it captures no model-layer economics and bears inference as COGS, visible in the $18.4 million rise in hosting costs. Its distinctive position is the regulated perimeter. |
| Products (share / barrier) |
|
|
|
| Long-horizon vote | +0.24 at weight 0.20 · swarm neutral Editorial prior, not backtested. | +0.38 at weight 0.20 · swarm neutral Editorial prior, not backtested. | +0.06 at weight 0.20 · swarm neutral Editorial prior, not backtested. |