market-map

Summary

Phase 2 market map for Mysterium Network, testing the Phase 1 working hypothesis that GoProxies is Mysterium's internal B2B brand, benchmarking the residential-proxy/agent-access competitive field from primary sources, and hunting for a nameable, reachable buyer pool for "unblockable access" demand. Every claim below is source-tagged with a confidence mark. Where evidence conflicts, both sides are shown.

Workstream 1 — GoProxies ↔ Mysterium relationship

The Phase 1 hypothesis does not survive contact with primary sources

Finding 1 — Mysterium's own documentation calls GoProxies a partner, not a subsidiary. The official Mysterium docs page states GoProxies is "our partner's ethical proxy network which leverages Mysterium's node network," and Mysterium's own /partners page lists GoProxies in a "Products Using Mysterium Technology" category alongside clearly-third-party products: Storj DCS, Polygon, Portals VPN, Avado, Nebra, DAppNode. [solid] — Source: https://docs.mysterium.network/about-mysterium and https://www.mysterium.network/partners (fetched 2026-07-06).

Finding 2 — A December 2023 press release treats them as two separate companies entering a partnership, not a rebrand or internal launch. Dateline VILNIUS, LITHUANIA; named contact Guoda Sulcaite ("Marketing manager," +370 641 62749, marketing@goproxies.com) issued from GoProxies' side, describing "GoProxies... has entered into a strategic partnership with Mysterium Nodes." The release's "About GoProxies" boilerplate describes it as an independent "premium proxy provider" with no mention of Mysterium ownership. [solid] — Source: https://www.accessnewswire.com/newsroom/en/computers-technology-and-internet/goproxies-partner-up-with-mysterium-nodes-796827 (dated 2023-12-07, fetched 2026-07-06). Corroborated by wire syndication on bignewsnetwork.com, clevelandstar.com, finance.yahoo.com — all same text, no independent added detail.

Finding 3 — No legal entity name appears anywhere in GoProxies' own ToS or Privacy Policy, which is itself notable. The ToS (/terms-and-conditions, v1.2, updated 27 May 2024) never states an operating company name or registration number — unusual for a ToS, and a genuine opacity gap. What it does establish: "These Terms shall be construed in accordance with the laws of the Republic of Lithuania... filed exclusively in the competent courts of the Republic of Lithuania" (§11.4). [solid, opacity itself] — Source: https://www.goproxies.com/terms-and-conditions (fetched 2026-07-06).

Finding 4 — The Privacy Policy on goproxies.com is written for a consumer VPN product (mentions "our VPN services," a "no logs pledge," mobile device/OS data collection, and Coingate crypto-payment processing) — not the B2B residential-proxy product the marketing site sells. This is either a template reused sloppily across a multi-product company, or evidence GoProxies operates (or shares infrastructure/legal boilerplate with) a consumer VPN arm too. [estimate — the mismatch is solid, the interpretation is not] — Source: https://www.goproxies.com/privacy-policy (fetched 2026-07-06).

Finding 5 — No Lithuanian registry hit for "GoProxies." Targeted searches against Okredo/rekvizitai/Scoris-style Lithuanian company databases returned zero matches for "GoProxies" as a company name. This means either: (a) it trades under a different legal name not surfaced by these searches, (b) it's incorporated outside Lithuania despite the ToS choosing Lithuanian law, or (c) search coverage simply missed it. Not resolved. [unknown] — Source: WebSearch queries against rekvizitai.lt/Okredo, 2026-07-06, zero direct hits.

Finding 6 — Mysterium itself runs through more than one legal entity, and has changed entities recently. Mysterium Network's core protocol is developed by BlockDev AG, Alpenstrasse 14, 6300 Zug, Switzerland [solid — Source: https://docs.mysterium.network, corroborated by Crunchbase/CB Insights entity listings]. Separately, as of January 1, 2025, MN Intelligence UAB (registered 1 Aludarių St., Vilnius, Lithuania, company code 307039286, founded 2024-12-03, director Ieva Matulaitienė, 13 employees as of mid-2026, 2025 revenue €1,132,969 with a net loss of €268,369) became the service provider for Mysterium VPN specifically. [solid] — Source: Mysterium VPN blog "Legal entity change" (referenced via search; direct fetch blocked by 403) and Lithuanian registry aggregators (Okredo/tax.lt/rekvi.lt/info.lt), cross-corroborating company code 307039286, 2026-07-06. Note: this is a consumer VPN entity, distinct from any GoProxies entity — no registry link was found tying MN Intelligence to GoProxies.

Finding 7 — Team size and independence signals. LinkedIn/Tracxn/RocketReach put GoProxies at roughly 6–50 employees (sources disagree: LinkedIn says 11–50, Tracxn's live count says 9, another aggregator says 6), Vilnius-based, with its own named Head of Sales (Vilius Kraučelis) and a 3-person sales team. Crunchbase lists GoProxies as unfunded, with an "obfuscated" founding date, no parent company field populated. [estimate — aggregator data, not primary, but multiple independent aggregators converge] — Sources: linkedin.com/company/goproxies, tracxn.com/d/companies/goproxies, rocketreach.co/goproxies-profile, crunchbase.com/organization/goproxies (all fetched via search 2026-07-06).

Verdict on the relationship

⚠ SUPERSEDED (2026-07-17): GoProxies confirmed IN-HOUSE (Lee, definitive, from inside). The "arm's-length partner" conclusion below is wrong — this pass over-read no visible public link as genuine independence. GoProxies is MN's own proxy product/brand; MN captures 100% of proxy revenue, no partner margin split. Kept below as the record of the outside-in pass and why it misfired (a deliberately-unstamped in-house brand is invisible to exactly this kind of external check).

The evidence argues against the ~80% "GoProxies is Mysterium's internal second brand" prior. What it supports instead: GoProxies is a genuine third-party proxy company (independent sales team, own marketing/PR voice, own — if unnamed — legal entity) that buys/leverages Mysterium's node network as an upstream ethical-IP supplier, formalized as a partnership in December 2023. This is a supplier relationship, structurally similar to how Bright Data or Oxylabs source some IPs from panel/SDK partners — not an org chart where GoProxies is Mysterium's storefront. It is plausible GoProxies is partly owned or seeded by people adjacent to Mysterium (Vilnius proxy-industry cluster is small and interconnected — see Workstream 4), but no document, filing, or registry record found in this pass establishes that. The GoProxies price sheet is a comparable, not proof of Mysterium's own price sheet.

This matters directly for the Phase 1 finding that "Mysterium's own domain has no B2B buyer path": if GoProxies is genuinely arm's length, then Mysterium is not quietly running B2B sales under a skin — it is leaving B2B monetization of its own node network to a third party's commercial relationship, which is a different (and arguably worse) strategic problem: Mysterium is the wholesale bandwidth supplier, not the customer-facing brand capturing margin.

GoProxies' actual pricing (read in full, since it's the closest primary proxy for what "buying from the Mysterium node network, packaged commercially" costs)

Home page headline tiers (from $/GB or $/IP, "80M ethically sourced IPs," 200+ locations):

Full rotating-residential price ladder (/pricing/rotating-residential-proxies-pricing, live 2026-07-06):

[solid] — Source: https://www.goproxies.com/pricing/rotating-residential-proxies-pricing, live page fetched 2026-07-06.

Positioning language: "ethically sourced," "lowest fraud scores in the industry," 99.99% uptime, unlimited concurrent sessions, dedicated account manager from the Micro tier up, 24/7 support via Slack/Telegram/email, no credit card required to start. This positioning — heavy emphasis on "ethically sourced" and "lowest fraud scores" — is the one place GoProxies' marketing voice echoes Mysterium's own "tamper-proof, ethical, decentralized" language, which is suggestive but not proof of common ownership. [solid on content, estimate on inference]

Workstream 2 — Competitor matrix (primary sources only)

Scale comparables (context for where Mysterium sits)

• Bright Data: ~$300M ARR in 2025 (up from $127.7M in 2024, >50% YoY growth), CEO projecting ~$400M ARR by mid-2026; most recent disclosed valuation $383.1M per PitchBook though recent reporting suggests >$1B. [solid — multiple converging sources: getlatka.com, calcalistech.com, arr.club, 2026-07-06 search]
• Oxylabs: $43.7M ARR (2025), ~397 employees, $131M disclosed valuation, built without any VC funding, same Tesonet venture-studio lineage as Nord Security/Surfshark/Decodo/Hostinger, HQ in the same Vilnius "Cyber City" building complex. [solid] getlatka.com, workwork.lt, corroborated across sources.
• Grass: pre-valuation ~$100M as of the Oct 2025 bridge round. [solid, per funding trackers cited above]
• Mysterium (Phase 1 finding, restated for contrast): MYST token market cap ~$2–3.6M. [carried over from Phase 1 — solid per Phase 1's own sourcing]

The gap between Mysterium's ~$2–3.6M token cap and Oxylabs' $131M valuation — a company in the same city, same non-VC bootstrapped-growth story, similar-vintage founding — is the single starkest number in this whole map.

Legal-clarity finding (important, cross-cutting)

Bright Data won two precedent-setting US federal rulings: Meta v. Bright Data (Meta's claims dismissed/abandoned, Jan–Feb 2024) and X Corp. v. Bright Data (dismissed May 2024, judge ruled X failed to state a claim over access to/copying of public data). Both rulings establish that scraping publicly accessible data via residential proxies is, at minimum, defensible under current US law when done at arm's length from the platform being scraped. [solid] — Sources: TechCrunch (2024-01-24, 2024-02-26), CNBC (2024-05-10), Calcalistech, Skadden/Proskauer/Quinn Emanuel law-firm client alerts, all converging on the same case outcomes.

This is directly relevant to Mysterium's "decentralized, ethically-sourced" positioning: the compliance/ethics story only matters commercially if buyers believe it reduces legal risk relative to unclear-provenance competitors — and Bright Data just proved in court that provenance opacity isn't fatal anyway, which undercuts "ethical sourcing" as a differentiator on legal-risk grounds alone (it may still matter on brand/PR grounds).

Workstream 3 — Demand-side pain evidence

Concrete, named evidence of agent-vs-anti-bot pain

1. GitHub issues, named and dated, on browser-use (13k+-star agent browser-automation framework): - Issue #2511, "Bug: set stealth=True but still blocked by Cloudflare's security," reporter samyScott, closed 2025-07-21. [solid] - Issue #1582, "Some websites not able to get past Cloudflare after Patchright update," reporter edwardysun, closed 2025-09-07. [solid] - Issue #356, "How do i bypass cloudflare captcha?" — long-running open issue. [solid] — Source: github.com/browser-use/browser-use/issues, fetched 2026-07-06.

2. Cross-framework acknowledgment of the same wall. Industry commentary (Futurum Group VP Mitch Ashley, quoted in trade press) states plainly: "AI agents paired with anti-detection scraping frameworks expose that identification-based access controls have a structural ceiling" — i.e., this is not a one-framework bug, it's structural to how agents present themselves to the web. [solid quote, estimate on how representative] — Source: browseract.com/blog (2026), cross-referenced against multiple 2026 "bypass Cloudflare" guides (scrapfly.io, webclaw.io, scrapfly.io) that all treat "AI agent gets blocked" as the default assumption needing a workaround chapter.

3. A regulatory tightening event with a hard date, working directly in Mysterium's favor as a demand driver. Cloudflare announced that starting September 15, 2026, new domains will by default block bots classified as "Training" or "Agent" on any page displaying ads (Search remains allowed). [solid, this is a forward-dated primary-source policy change] — Source: developers.cloudflare.com/bots/additional-configurations/block-ai-bots, cross-referenced via search 2026-07-06. This is a concrete, near-term catalyst: agent traffic that currently squeaks through will face a harder wall in about ten weeks from today (2026-07-06), which should increase willingness-to-pay for unblockable/residential access specifically among agent-builders.

4. Named companies actively building and funding "make agents look human" infrastructure in the last ~12 months: - Browserbase — $40M Series B, $300M valuation (June 2025); 50M browser sessions run in 2025 across 1,000+ customers; explicitly markets to "AI agent builders." [solid] - Anchor Browser — ships "Anchor Chromium" (stealth fork) and touts "Web-Bot-Auth Verified" status with Cloudflare specifically, i.e. its core pitch is passing Cloudflare's own verification scheme. Funding undisclosed. [solid on product claims, unknown on funding] - Firecrawl — $15M Series A (Aug 2025, led by Nexus Venture Partners), YC-backed, 350,000+ developers, 48,000+ GitHub stars, framed as "the context API to search, scrape, and interact with the web at scale" for LLM/agent use. [solid] - Oxylabs' own "Agent Skills" GitHub-distributed integration for Claude/Cursor/Copilot is itself evidence the giants see agent-builders as a named, worth-a-dedicated-product segment. [solid] - NetNut's per-model scraper products ("ChatGPT Scraper," "Perplexity Scraper," "Gemini Scraper") are further evidence the incumbents already treat "help my agent/LLM access data" as a distinct, monetizable SKU, not a side effect of generic proxy sales. [solid]

5. What's missing — no evidence found of a "≥30 nameable companies desperate in one vertical" pool. Search did not surface a dense cluster of named, distressed buyers (e.g., a specific list of 30+ startups all complaining about the same blocker) — what exists instead is a smaller number of well-funded infrastructure vendors (Browserbase, Anchor, Firecrawl, Steel.dev, Hyperbrowser) building solutions for a broad, diffuse population of agent-framework developers (browser-use, Stagehand, CrewAI, LangChain users) who show up in GitHub issues individually, not as an organized buyer segment. [negative finding, explicitly flagged as requested]

Named companies and roles, for reference

Verdict: which vertical concentrates the pain

The pain is most concentrated, and most commercially organized (i.e., already has vendors selling directly against it), in AI agent / LLM data-access tooling — specifically the sub-segment of teams building browser-automation agents (not simple API-based scrapers) that need to look like a real human browsing session end-to-end: fingerprint, IP reputation, session behavior, and CAPTCHA-solving together. This is narrower than "AI agents" broadly and narrower than "web scraping" broadly — it's the intersection. Evidence for concentration here specifically: three funded infra startups (Browserbase, Anchor, Firecrawl) built entirely around this intersection in the last 12–18 months, two proxy incumbents (Oxylabs, NetNut) shipped agent-specific SKUs in response, and Cloudflare is about to make the underlying problem structurally worse on a fixed date. It is a real, growing, and reachable vertical — but it is vendor-dense already (see Workstream 4) rather than a greenfield of 30 desperate unserved buyers.

Workstream 4 — Right to win

Where a $2–4M-cap, 25-person, decentralized, token-based player has genuine structural advantage

Cost structure of decentralized supply. Mysterium's node-operator model means IP acquisition cost is fundamentally different from Bright Data/Oxylabs' SDK-bundling or Grass's points-and-token consumer acquisition: nodes are paid in MYST for bandwidth already, and the network already exists. Marginal cost of serving another GB, once the node network is live, should be structurally lower than a competitor that has to keep paying SDK partners or consumer apps for opt-in bandwidth. This is a real potential advantage, but Phase 1 already found the company isn't capturing it commercially — the GoProxies relationship looks like Mysterium selling wholesale bandwidth to a customer-facing brand rather than capturing retail margin itself. [estimate — cost structure logic is sound, but no revenue-split data available to confirm Mysterium actually realizes the margin advantage]

Compliance/ethics story. Every single competitor in the matrix above (GoProxies, Bright Data, Oxylabs, Decodo, IPRoyal, NetNut) already claims "ethically sourced" / opt-in / consent-based IP acquisition. This is now table stakes, not a differentiator — and Bright Data's court wins show the market doesn't require it anyway. Mysterium being decentralized and open-protocol rather than a company claiming to be ethical is a genuine structural difference (verifiable on-chain / open-source vs. a marketing claim), but there is no evidence buyers currently pay a premium for that distinction. [estimate]

The Vilnius/Lithuania cluster. Real and valuable, but it's already fully occupied — Tesonet's venture studio (Oxylabs, Decodo/Smartproxy, Nord Security, Surfshark) sits in the same city, shares an office complex, and has multiple bootstrapped, non-VC-funded proxy/VPN companies already at $40–300M ARR scale using the exact playbook (consumer app SDK bandwidth → B2B proxy resale) Mysterium would need to compete on. GoProxies itself is Vilnius-based. Mysterium is not entering an empty cluster — it is the smallest, least-capitalized player in a cluster that already has category leaders. This cuts against, not for, entering on cluster-proximity alone.

Niche verticals the giants ignore. The giants (Bright Data, Oxylabs, Decodo, NetNut) have already built agent-specific SKUs (Agent Skills, model-scrapers) and are moving fast into the "AI agent access" niche identified in Workstream 3 — so "the giants ignore this" is not currently true and closing as a window. The more plausible niche-the-giants-ignore is the decentralized/crypto-native developer: teams building on-chain, DePIN-adjacent, or privacy-maximalist agent infrastructure who want a protocol, not a vendor relationship — pay-per-use via token/crypto rails without a corporate intermediary, verifiable node provenance, no SOC2-theater required because the trust model is different (cryptographic/economic, not audit-based). This is genuinely underserved: Grass is the closest comparable and is consumer/AI-training-focused, not agent-access-focused; none of Bright Data/Oxylabs/Decodo/NetNut offer a crypto-native, no-KYC, protocol-level buy path.

Where entering is suicide

Competing head-on for generic "residential proxy for scraping/BI" B2B customers against Bright Data ($300M ARR, court-tested legal position, 14 of top-20 LLM labs as customers) or Oxylabs (same city, same non-VC bootstrap story, already at $131M valuation with 397 employees) with a 25-person team and a $2–4M token cap is not a fight — it's a rounding error entering a mature, consolidating market with entrenched sales motions, ISO/SOC certification moats, and incumbents who've already built the exact AI-agent SKU Mysterium would pitch as its wedge. Racing Browserbase/Anchor/Firecrawl on "agent browser infrastructure" is worse: those are freshly-funded ($15–40M rounds), purpose-built, developer-experience-obsessed products competing on a completely different axis (session/browser-hours, not just IP supply) that Mysterium's node network alone doesn't answer.

The right-to-win answer, three sentences

Mysterium's only defensible wedge is protocol-level, crypto-native access for decentralized/agent-infrastructure builders who explicitly don't want a vendor relationship — pay-per-GB in MYST or stablecoin, no KYC, no SOC2 theater, verifiable node provenance — a niche where Grass is the only real comparable and none of the proxy incumbents currently compete. Winning there requires abandoning any ambition to out-sell Bright Data/Oxylabs/GoProxies on generic B2B proxy terms, and instead formalizing (or acquiring outright) the GoProxies-style commercial layer so Mysterium itself — not an arm's-length partner — captures the retail margin on its own node network. Without that ownership move, Mysterium remains structurally what Workstream 1 suggests it already is: a wholesale bandwidth supplier to someone else's storefront, which is not a path to break-even at any interesting speed.

Kill/confirm verdicts

H1 — "Mysterium B2B is price-viable but invisible." Partially confirmed, materially revised. The pricing IS viable and comparable to the market (GoProxies' $1.40–2.59/GB sits squarely mid-pack against NetNut, Decodo, IPRoyal). But the "invisible" framing undersells the real problem: it's not that Mysterium's own B2B offer is hidden behind a rebrand — it's that Mysterium may not have a B2B commercial offer of its own at all; the commercial capture appears to sit with a separate company (GoProxies) via a partnership, not a skin. If true, this is worse than "invisible pricing" — it's "no direct pricing to be invisible," because Mysterium isn't the seller of record. [estimate — confirmed on pricing-viability, revised on the mechanism]

H2 — "Desperate agent-access buyer pool ≥30 nameable companies in one vertical." Killed as stated, but a real (smaller, differently-shaped) opportunity exists nearby. No evidence surfaced of 30+ nameable, desperate companies in one vertical begging for access. What exists instead: a well-evidenced, real, and growing pain point (agents blocked by anti-bot systems, worsening on a fixed Sept 15 2026 Cloudflare deadline) that is already being served by 3+ freshly-funded infrastructure vendors (Browserbase, Anchor, Firecrawl) and 2+ incumbent proxy companies with dedicated SKUs (Oxylabs Agent Skills, NetNut model-scrapers). The vertical is real; the "≥30 nameable, unserved" framing is not supported — it's a vendor-crowded, VC-funded space, not a greenfield. [killed as originally framed]

H3 — "Pricing opacity costs deals." Confirmed, and worse than the hypothesis implies. GoProxies' ToS names no legal entity at all — an actual, verifiable opacity gap in a document whose entire purpose is legal clarity. Compare to Oxylabs and Decodo, which lead with certification badges (ISO/SOC2) precisely because enterprise procurement teams require them before signing. A B2B buyer doing basic diligence on GoProxies today cannot even confirm who they're contracting with from the public-facing legal pages alone. This is very likely costing larger/more risk-averse deals specifically, though no direct "we walked away because of this" evidence was found (would require buyer-side interviews, out of scope for this desk pass). [confirmed on the documentary gap; the deal-loss causal claim remains estimate/unknown for lack of buyer testimony]