Vendor Sites Win 3 of 6 Cybersecurity AI Citation Shapes

Today's Machine Relations Index publishes all six buyer-question shapes for two categories a large-company buyer researches back to back: AI Infrastructure and Cybersecurity. In AI Infrastructure, a vendor's own domain holds the #1 citation slot in zero of the six segments. In Cybersecurity, a vendor's own domain holds #1 in three of six, and vendor-owned domains fill more than twice the top-nine share.

Canonical URL: https://authoritytech.io/curated/vendor-sites-win-cybersecurity-not-ai-infrastructure-2026 Published: 2026-09-26 Author: Jaxon Parrott Tags: Afternoon Brief, AI Search & Discovery, Measurement

If your content plan assumes "publish it on our own site and the AI will cite us," that plan's odds depend entirely on which category you're in — and this pair of categories shows the two ends of the range.

What today's release publishes

Release mri_score_v2.0+2026-09-26+2b779408cfda, generated 2026-09-26, observation window 2026-05-10 through 2026-09-26. Read from the Machine Relations Index, which classifies every cited domain into a source-role class: editorial publication, vendor-owned, market and company database, academic and government, community and social, analyst and consulting research, search or media platform, wire and press-release distribution, or an unclassified "other observed source" bucket this brief treats the same way the Index's own category studies do — read for its winners but not compared as a class.

AI Infrastructure and Cybersecurity both have all six buyer-question segments published and clear of the evidence floor (at least 10 observed runs across at least 7 distinct dates): AI Infrastructure at 101 to 136 observed runs per segment, Cybersecurity at 105 to 131.

The #1 slot, six shapes each

Buyer question	AI Infrastructure #1	Source role	Cybersecurity #1	Source role
Best tools	Medium, 42.06%	Editorial publication	Huntress, 53.33%	Vendor-owned
How buyers choose	Arxiv.org, 28.68%	Academic and government	Microsoft, 21.37%	Vendor-owned
Is it worth it	Reddit, 25.93%	Community and social	Palo Alto Networks, 28.24%	Vendor-owned
Problem-first research	Reddit, 18.38%	Community and social	GitHub, 18.40%	Community and social
Top lists	Spheron.network, 37.04%	Other observed source	Hoxhunt, 28.46%	Other observed source
Head to head	YouTube, 33.33%	Search or media platform	Exabeam, 26.17%	Other observed source

Zero of the six AI Infrastructure #1 slots is a vendor's own domain. Three of the six Cybersecurity #1 slots are: Huntress on best tools, Microsoft on how buyers choose, Palo Alto Networks on is it worth it — the three shapes closest to an active purchase decision.

The gap holds past the #1 slot

Counting every top-nine row across the six segments in each category, not just the leader, sharpens the same split. AI Infrastructure's 54 top-nine slots include seven vendor-owned domains (Nvidia twice, Atlan, OpenAI, Databricks, IBM, Unstructured.io) — 13% of the rows. Cybersecurity's 54 top-nine slots include 17 vendor-owned domains (SentinelOne appears four times across shapes, plus Cynet, Palo Alto Networks, Microsoft, Rapid7 and CrowdStrike) — 31% of the rows, better than double.

What fills the AI Infrastructure gap instead: Reddit, Arxiv, Medium, YouTube and Dev.to appear a combined 15 times across the 54 slots. Academic and community sources carry the category's buyer questions; vendor domains barely register outside "how buyers choose" and "head to head," and never lead either.

Gartner appears once across both categories' 108 top-nine slots — #8 on cybersecurity top lists at 16.15% — which matches Machine Relations' portfolio-wide finding that no analyst-and-consulting domain leads any of 88 published segments.

What this changes for a marketing plan

If you sell cybersecurity, your own domain is a live citation channel for the exact questions that decide a purchase. Huntress's product pages, Microsoft's and Palo Alto Networks' documentation and comparison content are already what four of six answer engines cite when a buyer asks "how do I choose" or "is it worth it." A content plan that routes budget away from the vendor's own site toward earned placements alone is optimizing against what these three segments show is already working.

If you sell AI infrastructure or developer tooling, the same plan needs a different center of gravity. No AI Infrastructure buyer-question segment cites a vendor's own domain first, and only two of six cite one at all in the top nine. The segments are won by Arxiv papers, Reddit threads, Medium posts and YouTube walkthroughs — third-party and community explanation, not vendor product pages. A vendor betting its AI-citation strategy on its own blog and docs is betting against what this category's six segments currently show.

Both readings hold only for the buyer-question shapes measured here, not for the news-driven segment in either category, and only as long as this release's ranking holds — check the current release before repeating either number.

Method

Data: Machine Relations Index v2, release mri_score_v2.0+2026-09-26+2b779408cfda, window 2026-05-10 to 2026-09-26, six answer engines (ChatGPT, Claude, Gemini, Google AI Mode, Google AI Overviews, Perplexity). Segments require at least 10 observed answer runs across at least 7 distinct run dates before publishing a rate. Figures above are read directly from the published AI Infrastructure and Cybersecurity category-and-shape pages; source-role labels are the Index's own declared taxonomy. AuthorityTech applies Machine Relations' research; it is not our product.