Ask it what you'd ask a research team
Built for the analyst's second question
Dashboards answer the questions someone pre-modelled. Hosting markets rarely fit a clean grid — the question that matters is usually the second or third: Is that real hosting or parked inventory? Did ownership change without an infrastructure migration? Is this growth market-wide or operator-specific? HostingBrain is built for that follow-up: you ask in your own words and keep drilling, each answer carrying its denominator, attribution layer and caveats.
A typical chain, in one conversation:
- Which groups are consolidating hosting in Denmark, and how fast?
- For each — full control-plane book vs confirmed-active — which number am I looking at?
- Which of those operators run their own infrastructure, and which resell?
- Which ownership links are press-ledger-validated, and which are inferred?
- What caveats apply before this goes in a memo?
Built for outside-in diligence and market mapping — PE & family-office screening, advisor diligence, operator competitive intelligence, post-acquisition integration visibility, and product/strategy trend tracking — inside the AI assistant your team already uses, not one more dashboard to log into.
Why domain counts mislead
The same "hosting footprint" looks radically different depending on the denominator. Illustrative archetypes below — the named, ranked table (and per-group ownership, book quality and dynamics) is in the product, under acceptable-use terms.
| operator archetype | full book | confirmed-active | live share |
|---|---|---|---|
| European platform consolidator | 1–3M | 0.7–2M | 60–80% |
| Regional integrated hoster | 100k–500k | 60k–350k | 50–85% |
| Portfolio-heavy operator | 500k–2M | 50k–300k | 10–30% |
A domain book and an active-hosting footprint are different things, and the ratio varies widely by operator type — which is why raw domain counts overstate some and understate others. Confirmed-active is a floor: operators that serve their customers on third-party cloud can be undercounted by infrastructure attribution, so we state that as a caveat rather than mistake it for a low hosting business. Snapshot 2026-07-05.
See it answer
Not mockups — actual product output from the current snapshot, restyled for the page. Market structure is shown openly; forward-looking and provider-level intelligence is redacted here and named in the product.
| year | acquirer group | target | category | source |
|---|---|---|---|---|
| 2020 | group.one | Hostnet | hosting / domains | company release |
| 2024 | team.blue | Loopia Group | hosting | company release |
| 2024 | Miss Group | Domeneshop | domains / registrar | industry press |
| 2025 | World Host Group | A2 Hosting | hosting | PR Newswire |
| 2025 | team.blue | AssoConnect | SaaS | company release |
Why this exists
HostingBrain began in 2025 as an internal research project: systematically observing Europe's web-provisioned domain population to understand a quietly consolidating industry — who actually hosts the continent's businesses, who is buying whom, and what the books being traded really contain. The dataset grew into the substrate for private analyst briefings on hosting M&A and market structure.
When we decided to open it up, we made a deliberate choice. The analysts we worked with weren't reading dashboards — they were already doing their analysis in conversation with AI assistants. So rather than shipping another BI tool or dumping raw data, we built the dataset into the place where the work already happens: an AI-native connector that answers questions the way an informed colleague would — with the date, the denominator and the caveats attached. Expert sessions don't scale; giving the analyst the expert's dataset, inside their own workflow, does.
As part of opening external access, the project is being spun out into its own company — HostingBrain ApS (under formation, Denmark).
Data an analyst can defend
Every answer carries its as-of date, its denominator, and its caveats. Coverage is reported as a spectrum, and every number states which end it sits on:
- 285M+ web-provisioned — the outer bound: domains observed with web/DNS provisioning (a ceiling; includes registrar-parked and placeholder domains).
- 126M+ web-active — the working-web middle bound: resolves to real hosting infrastructure, with parking and registrar/forwarding placeholder IPs excluded (identified by reverse-DNS operator fingerprints, not just IP location — so a placeholder on a major cloud is still excluded).
- 35M+ confirmed-active business domains — the defensible floor: verified operating email, live content or platform provisioning; parking, registrar placeholders and dead pages excluded. This is a domain classification, not a customer, company or legal-entity count.
We report the whole spectrum and label which end each number sits on. A domain that merely resolves to a shared parking or registrar/forwarding placeholder server is not counted as web-active — even when that server is on a major cloud — and headline business claims use the confirmed-active floor, never the ceiling.
European depth, global benchmark. Our ownership, consolidation and firmographic analysis is European-deep and calibrated against public registry totals — but we don't just rank European ccTLD books; we benchmark European operators against the whole global web-provisioned universe, so every European figure comes with global context. Control-plane attribution is separated from true hosting, calibrated signals from direction-only. When we don't know, the answer says so — that's the product.
| layer | evidence | confidence |
|---|---|---|
| Registrar / control plane | nameserver + registrar pattern | Medium |
| Web serving | origin IP + reverse-DNS + TLS / platform fingerprint | High |
| Ownership | press-validated acquisition ledger | High |
| Business activity | live content + MX + platform provisioning | Medium |
| Caveat | brand retained separate infra, or serves on third-party cloud | stated |
External scale check: the global registered-domain universe is materially larger than our observed web-provisioned population, and web-server surveys count hostnames and sites differently from our domain-level classification — so our figures are not directly comparable to registry or survey totals, by design.
How it works
HostingBrain is built on the open Model Context Protocol (MCP) — so it lives inside the AI assistant you already use, not behind another login. No vendor lock-in: the same endpoint works in Claude, Claude Code, Cursor, and any MCP-compatible client.
Claude (web, desktop, iOS/iPad): Settings → Connectors → Add custom connector →
https://mcp.hostingbrain.ai/mcp
A connect window opens — paste your access key once, done. In Claude Code, the API, Cursor, or any other MCP client, attach the key as a bearer header directly. ChatGPT connector support: on the roadmap.
Access
We're in pre-launch: the free tier is open, paid tiers are invite-only while we onboard early users personally.
Free
- Market digitization index
- Coverage & methodology calibration
- 30 queries/day
Pro
- Named-provider market structure
- Book health & peer comparison
- Consolidation landscape & SaaS-shift analysis
Analyst
- Everything in Pro
- Operator dossiers & ownership evidence
- Market mapping & footprint dynamics
- 14-day trial
Deal-Pass
- Full access for one mandate
- 6–8 weeks, no subscription
- Expense it to the deal
Free key
Request an invite
FAQ
What this is not
Sophisticated buyers trust products that state their boundaries. HostingBrain is not:
- a revenue estimate, or a substitute for a company's internal metrics;
- a full customer list, or a legal register of ownership;
- a claim that DNS or ASN attribution equals hosting — we separate control-plane, serving, and ownership;
- perfect on resellers, white-label operators, CDNs, agencies or parked portfolios — those are stated as caveats, not hidden.
When we don't know, the answer says so. That's the trust anchor of the whole product.
Trust
Research confidentiality: your query subjects are never visible to other customers, never sold, and query parameters are reduced to theme-level aggregates after 90 days. Honesty over polish: if a number is a floor, a ceiling, or direction-only, the answer says so. Errata: if we materially correct a served figure, subscribers are notified. Details: security · terms · privacy.