insights.trendforce.com APIinsights.trendforce.com ↗
Retrieve TrendForce Insights articles and post listings via API. Access full article text sections, figures, tags, authors, and metadata for AI chip and semiconductor analysis.
curl -X GET 'https://api.parse.bot/scraper/fbe93ce5-a6bd-4c39-8e93-3a714d6de4c4/list_posts' \ -H 'X-API-Key: $PARSE_API_KEY'
List and search posts from TrendForce Insights. Supports sorting by newest or most popular, keyword search, and pagination via offset/limit.
| Param | Type | Description |
|---|---|---|
| sort | string | Sort order. Accepts exactly: new, top. |
| limit | integer | Maximum number of posts to return per request. |
| offset | integer | Pagination offset (number of posts to skip). |
| search | string | Search query to filter posts by keyword (e.g. 'NVIDIA', 'AI chip'). Empty string returns all posts. |
{
"type": "object",
"fields": {
"count": "integer",
"posts": "array of post summary objects",
"offset": "integer"
},
"sample": {
"count": 1,
"posts": [
{
"id": 199556891,
"slug": "ai-inference-chip-architecture",
"tags": [
"Computing-in-Memory (CIM)",
"AI Inference",
"Cerebras"
],
"title": "The Inference Economy Arrives: AI Chip Rules Are Being Rewritten",
"authors": [
"TrendForce"
],
"subtitle": "When Tokens Cost Money, Architecture Becomes Strategy",
"post_date": "2026-05-29T04:01:43.840Z",
"wordcount": 1603,
"cover_image": "https://substack-post-media.s3.amazonaws.com/public/images/d9638e76-c339-4563-84c8-3c04e92c2c54_1731x909.png",
"description": "When Tokens Cost Money, Architecture Becomes Strategy",
"canonical_url": "https://insights.trendforce.com/p/ai-inference-chip-architecture",
"comment_count": 0,
"reaction_count": 4
}
],
"offset": 0
}
}About the insights.trendforce.com API
The TrendForce Insights API exposes 2 endpoints for accessing semiconductor and AI industry analysis from insights.trendforce.com. The list_posts endpoint returns paginated post summaries with keyword search and sort controls, while get_post delivers full article content broken into text_sections and figures — including alt text, captions, tags, authors, subtitle, wordcount, and publication date — structured for direct LLM ingestion.
Post Listings
The list_posts endpoint returns a count, a posts array of summary objects, and the current offset. You can sort results by new (most recent) or top (most popular), paginate using limit and offset, and filter by keyword with the search parameter — for example, passing 'NVIDIA' or 'HBM' scopes results to matching articles. Leaving search as an empty string returns all posts.
Full Article Content
The get_post endpoint takes a slug — either copied from a URL path like ai-inference-chip-architecture or pulled from a list_posts result — and returns the complete article. The response includes title, subtitle, authors, post_date, wordcount, tags, and cover_image as metadata fields. Article body content is split into text_sections (paragraphs, headings, list items, and blockquotes) and figures (each with src, alt, and caption). This split makes it straightforward to feed article text and image descriptions into downstream language model pipelines without additional parsing.
Coverage and Scope
TrendForce Insights focuses on semiconductors, memory, display panels, AI chips, and related supply chain topics. Articles are authored by TrendForce analysts and cover market share estimates, technology roadmaps, and supply/demand forecasts. The tags field on each post signals the topic cluster, and authors identifies the analyst responsible — useful for filtering by subject-matter specialization.
- Monitor new TrendForce analyst reports on AI inference chips by polling
list_postswithsort=newand a relevantsearchkeyword - Build a semiconductor news digest by fetching
text_sectionsfromget_postand summarizing with an LLM - Index TrendForce articles by
tagsto track coverage of specific topics like HBM, TSMC, or edge AI over time - Extract
figureswith captions from chip architecture articles to populate a visual reference database - Aggregate
authorsandpost_datefields to analyze publishing cadence and analyst focus areas - Feed
wordcountandsubtitlefields into a content scoring pipeline to prioritize long-form research for deeper analysis - Search for competitor mentions across articles using the
searchparameter inlist_poststo track how frequently specific companies appear
| Tier | Price | Credits/month | Rate limit |
|---|---|---|---|
| Free | $0/mo | 100 | 5 req/min |
| Hobby | $30/mo | 1,000 | 20 req/min |
| Developer | $100/mo | 5,000 | 250 req/min |
One credit = one API call regardless of which marketplace API you call. Exceeding the rate limit returns a 429 response. Authenticate with the X-API-Key header.
Does TrendForce offer an official developer API for insights.trendforce.com?+
What does `get_post` return beyond the article text?+
id, slug, title, subtitle, authors (array), tags (array), post_date, wordcount, and cover_image as metadata. The body is divided into text_sections — covering paragraphs, headings, list items, and blockquotes — and figures, each containing src, alt, and caption. Premium or paywalled content that requires a TrendForce subscription is not exposed by the endpoint.Can I filter `list_posts` results by tag, author, or date range?+
search parameter and sorting by new or top. Filtering by specific tag, author, or date range is not available. You can fork this API on Parse and revise it to add those filter parameters.Is there a way to get comment counts, view counts, or social engagement metrics for each post?+
list_posts response returns post summaries and the get_post response covers article content and metadata; neither exposes engagement metrics such as comment counts or view counts. You can fork this API on Parse and revise it to add an engagement data endpoint if that surface is accessible.How does pagination work in `list_posts`?+
limit to set the number of posts per request and offset to skip a number of posts from the beginning of the result set. The response always includes count (total matching posts) so you can calculate how many pages remain. There is no cursor-based pagination — only numeric offset.