Vendors
86 articlesCompany profiles that accumulate across editions. Every claim carries a source link and a checked date.
An open archive of COMPUTEX exhibitors, products and editions.
Every fact carries a source link. What the organiser has not published stays blank.
Three rules decide what gets written here. They are also the reason an AI engine has any reason to quote this site rather than a vendor brochure.
Every field on a vendor page comes from the official exhibitor directory, pulled by a script, with no language model rewriting anything in between. Each row carries the source URL and the date it was checked. A vendor can send a pull request to correct their own facts, as long as the correction carries a source too.
Where a company sits in the supply chain, what changed since last year, how one edition compares to the last: that is curation, and only neutral editors write it. Vendors can dispute it in a pull request; they cannot edit it directly. That boundary is the design itself.
When the official directory does not publish a field, the page leaves it empty. Filling it in takes a verifiable source, not a plausible guess. A page that looks complete is worth nothing if a single row of it was invented.
Not sure where to start? Roll the dice and see what is in the archive.
Every fact carries a source link and the date it was last checked
Plain Markdown twins, llms.txt, JSON-LD and an MCP server, all public
One file per company, accumulating across years instead of resetting every show
Every page can be claimed and corrected through a pull request
Counted from the files themselves at build time. Nothing here is a rounded-up claim.
Counted from the files, not estimated. Empty means empty.
One page per exhibiting company, extracted from the official directory: venue, booth number, official site, product tags, and the full year-by-year exhibiting record.
Longest COMPUTEX records on file
Nothing here yet. This section stays empty until there are facts to put in it.
One page per edition: dates, halls, themes, and what was actually announced.
Nothing here yet. This section stays empty until there are facts to put in it.
Page counts are what actually exists on disk, not what the directory contains. Whenever the exhibitor count in the header and the page count here diverge, that gap is how much of the directory has not been turned into pages yet, and it is left visible on purpose.
Company profiles that accumulate across editions. Every claim carries a source link and a checked date.
Products and technologies shown at COMPUTEX, with specifications traced to the manufacturer.
Each year of the show: dates, venues, themes and what actually changed.
Editorial analysis of the Taiwan AI hardware supply chain. Written by neutral editors, not by vendors.
Every change to this archive is a public commit
fix(data): 第十三份寫死的母體分類表,以及它那個綠色的失敗訊息
fix(privacy): 停止使用母體的 Google Analytics 與 justfont 帳號
feat(content): 86 家廠商的英文頁,站上 115 → 201 頁
chore: 刪掉打不到的母體資產與死 CSS,dist 45M → 32M
fix(content): cross-domain: /mcp 教訪客安裝的是母體的 npm 套件
If your company is in here, the page is yours to correct. Facts change through pull requests, and every correction needs a source. If you would rather the page came down, the same route works.