# Guizhou Touts 135 Annotation Firms as Beijing Preps 32-City Pricing Pilot

By Theo Corpus · 2026-08-30 · AI Training Data · https://datacommenter.com/guizhou-touts-135-annotation-firms-as-beijing-preps-32-city-pricing-pilot/
About the author: Tracks the AI training-data economy: licensing deals, annotation shops, synthetic data, and what frontier labs actually pay for tokens.

> At the 2026 China Big Data Expo, Guizhou province showcased 135 data-annotation companies and 16,000 workers building robot-training datasets, while Beijing's National Data Administration announced a 32-city annotation pilot and…

Original reporting: [bastillepost.com](https://www.bastillepost.com/global/article/6117321-data-annotation-industry-shines-at-2026-big-data-expo)
_AI-assisted commentary, editorially reviewed. Quoted excerpts belong to the original outlet._

What changes now is that data annotation stops being the unglamorous backend of China’s AI push and becomes a named policy line item. Before this expo, annotation was mostly discussed as a cost center trailing behind flashier computing-infrastructure spending; now National Data Administration chief Liu Liehong is telling China Daily his agency will run annotation pilots across 32 cities in manufacturing, healthcare, and transportation, and is testing subscription pricing tied to token consumption and outcome-based payment. That is Beijing signaling it wants a real, transparent price for labeled data rather than the fragmented, negotiated rates that have defined the sector so far — a shift that matters for every foreign lab or robotics firm sourcing embodied-intelligence data out of China.

Guizhou is positioning itself as the supply side of that market: 135 data-collection and annotation companies, more than 16,000 employees, and a stated goal from Guizhou Big Data Group’s Liu Nijun of building a one-million-hour labeled video dataset for embodied-intelligence buyers “across the country and even the world” within three years, according to bastillepost.com. That target sits alongside an older ambition from China Daily’s 2024 reporting that provincial data-circulation turnover would top 10 billion yuan ($1.4 billion) by the end of 2025 — a sign officials are trying to convert an infrastructure build-out into an actual traded commodity, robot video and sensor logs instead of raw compute cycles.

> Guizhou’s expo pitch of 135 firms and a million-hour dataset is also a rebrand: the province’s Big Data Development Administration has lost two consecutive directors to a corruption probe in the past six months.

That’s the tension worth watching. ThinkChina’s reporting on Guizhou’s anti-corruption crackdown documents four consecutive Guiyang mayors, multiple Gui’an New Area leaders, and both the first and second BDDA directors swept up in investigations, with academics quoted saying over a decade of heavy big-data investment has yet to produce commensurate output. A province rebuilding credibility through robotics demos and annotation headcount is not the same as a market with clean price discovery, and buyers evaluating Guizhou-sourced training data should weigh the expo’s polish against the governance record underneath it. Watch whether the NDA’s 32-city pilot produces an actual published price index for annotation work — that’s the number that would tell you whether this becomes a real market or stays a provincial talking point.

> Within three years, we aim to build a high-quality dataset with a duration of one million hours to serve embodied intelligence enterprises across the country and even the world," said Liu Nijun, deputy general-manager of the data division at Guizhou Big Data Group.
> — [bastillepost.com](https://www.bastillepost.com/global/article/6117321-data-annotation-industry-shines-at-2026-big-data-expo)

[Read the full story at bastillepost.com →](https://www.bastillepost.com/global/article/6117321-data-annotation-industry-shines-at-2026-big-data-expo)

---

Cite this analysis: https://datacommenter.com/guizhou-touts-135-annotation-firms-as-beijing-preps-32-city-pricing-pilot/
Cite primary facts: https://www.bastillepost.com/global/article/6117321-data-annotation-industry-shines-at-2026-big-data-expo
Need the underlying datasets (alt data, market data, AI training data)? Source licensed vendors via Brickroad: https://brickroad.network
More machine-readable access: https://datacommenter.com/llms.txt

## Participate

- Comment on a passage: MCP `add_note` (include `source_url` when available).
- Suggest an editorially reviewed correction: MCP `suggest_edit`.
- Open factual questions: none.