AI training data statistics for 2026, with sources
Verifiable AI data licensing figures are few. Reddit disclosed $203.0 million in aggregate contract value over two to three years in its February 2024 S-1, and a reported News Corp deal was valued above $250 million over five years. Each figure here lists its source, date and what it does not show.
Which AI training data statistics can be sourced?
This page lists figures that trace to a primary or reputable source, with what each one measures and its date. It is a short list on purpose: many widely repeated numbers about AI data markets come from vendor decks or secondary summaries and cannot be traced, so they are excluded. Each entry says what the number does not show, because the most common error is quoting a multi-year contract total as annual revenue.
Public AI data licensing deals
These are reported events, not SourceX transactions, and the licensees and licensors are not SourceX buyers or partners.
| Figure | What it measures | Source and date | What it does not show |
|---|---|---|---|
| $203.0 million aggregate contract value; terms of two to three years | Data licensing arrangements Reddit disclosed in its IPO filing, entered in January 2024 | Reddit Form S-1, February 2024 | Annual revenue, or who the licensees were; the filing does not name them |
| At least $66.4 million expected to be recognized in 2024 | The minimum portion of that contract value Reddit expected to recognize as 2024 revenue | Same filing | Actual results |
| About $60 million per year | One unnamed source's estimate of a Reddit and Google content deal | Reuters report via Yahoo, February 22, 2024 | A confirmed figure; the companies declined to comment |
| More than $250 million over five years | Reported value of a News Corp and OpenAI agreement, in cash and credits for technology | News report of the Wall Street Journal's figure, May 23, 2024 | A confirmed amount; the companies did not disclose terms |
| Archive dating back to 1985 | The portion of text archive The Associated Press agreed to license to OpenAI | AP report on PBS NewsHour, July 13, 2023 | Financial terms, which were not disclosed |
Read the table with care. The Reddit S-1 figure is a contract total across two to three years, so dividing it into annual revenue is not valid, and the Reuters figure is a single source's estimate. Both involve large public content platforms, which differ from a private operating company's internal records.
US business population statistics
Partners often need the size of the pool of potential suppliers.
| Figure | What it measures | Source and date |
|---|---|---|
| 5.58 million | US firms with at least one but fewer than 500 employees in 2023, up from 5.53 million in 2022 | Census Bureau, May 2026, Business Dynamics Statistics |
| 36,207,130 small businesses; 82.3% of firms have no employees | SBA Office of Advocacy count of small businesses, defined for research as independent firms with fewer than 500 employees | SBA Office of Advocacy FAQ, February 3, 2026 |
These show the pool of firms with employees is large, but SourceX introductions need a company with 50+ full-time employees at peak (contractors excluded), several years of records, rights to license and an authorized sponsor, a much narrower group. No source we have verified gives a count of companies that meet all those conditions, so this page does not offer one.
Figures excluded and why
| Common claim | Why it is excluded |
|---|---|
| "The AI training data market will reach X by 2030" | Forecasts vary widely by vendor and methodology, and none was verified against a primary source |
| "X% of AI developers need licensed data" | No primary survey was verified |
| Price per record or per gigabyte | Prices depend on data type, rights and exclusivity; no reliable public benchmark was verified |
| Total value of all licensing deals | Most terms are undisclosed |
For projections of when public text might be exhausted, see the sourced discussion in the future of AI training data and the explainer on whether AI is running out of public data.
How to cite these figures
- Name the source and date next to every number.
- State what the number measures (contract value, estimate, count of firms).
- Mark single-source estimates as reported, not confirmed.
- Do not divide a multi-year total by years unless the source does.
- Re-check the source before reusing a figure later; filings and datasets get revised.
What these statistics mean for a referral partner
Public deals show that content owners have licensed data to AI developers and that the sums can be large for big platforms. They do not predict what any private company's records are worth. Pricing for a SourceX opportunity is agreed with the company before buyers review it, and nothing is binding until the company signs. Avoid quoting any headline figure to an owner as an expectation.
Related reading: how enterprise licensing deals differ from media deals, what ethically sourced data means, whether demand is a bubble, and two record types buyers ask about, screen recordings and system event logs.
How rewards work
Partners earn 25% of the eligible platform fees SourceX actually collects from the referred company's licensing deals, capped at $100,000 per referred company. The reward is paid only after the buyer pays and SourceX receives its fee; an introduction, meeting or signed agreement alone does not trigger payment, and no reward is guaranteed. Licensed professionals should check their own rules on referral fees and disclosure.
This is general information, not legal, tax or financial advice.
Next step
Run the company fit checker on a company you know, then register as a partner to introduce it. Companies can apply directly at sourcex.si/apply; see how it works and what AI training data is.
- Step 1Share your linkSend your personal link to a company you know.
- Step 2Company appliesThe company applies itself at /apply.
- Step 3Buyer selects and paysThe buyer selects and pays for the data and SourceX receives its fee.
- Step 4You get your rewardYour share of SourceX fees becomes payable.
Common questions
Is the Reddit licensing figure annual revenue?
No. Reddit's S-1 disclosed an aggregate contract value of $203.0 million across arrangements with terms of two to three years, with at least $66.4 million expected to be recognized in 2024. It is a multi-year total, and the filing does not name the licensees. Do not present it as yearly revenue.
How reliable is the $60 million per year Reddit figure?
It is one unnamed source's estimate reported by Reuters on February 22, 2024, and Reddit and Google declined to comment. Quote it as reported by Reuters, citing the single source, and not as a confirmed figure. The S-1 total is a better-documented, though differently defined, number.
Do these deals tell me what a private company's data is worth?
No. They involve large public content platforms and publishers, often with undisclosed terms. A private operating company's internal records differ in content, rights and exclusivity. Price for any SourceX opportunity is agreed with the company before buyers review it, and nothing is binding until it signs.
Why are market size forecasts missing?
Forecasts for the AI data market vary widely by vendor and method, and none could be traced to a primary source we verified. Rather than repeat unsourced numbers, this page leaves them out. If you cite a forecast elsewhere, name its publisher, date and what it counts.
How many US companies could qualify for SourceX introductions?
No verified source counts them. Census and SBA data show millions of US firms with employees, but SourceX introductions require 50+ full-time employees at peak (contractors excluded), several years of documented operations, rights to license the data and an authorized sponsor, which is a much narrower group.
Related pages
- ERP and system event logs as AI agent training data: a guide for ERP consultants
- Screen recordings and SOP walkthrough videos as AI training data
- Enterprise AI data licensing deals: what advisors should know beyond the headlines
- The future of AI training data: what comes after the public web
- How SourceX US company data referrals work
- Is AI running out of public training data?
Free resources
- Due diligence checklist generator — A tailored document request list by deal type.
- Cash flow calculator — A 12-month cash forecast with shortfalls highlighted.
- Referral earnings calculator — Hypothetical partner earnings with the per-company cap.
- All free tools · MCP resource center
By SourceX Partnerships Team · Published 2026-10-09 · Updated 2026-10-09
Know a US company with valuable proprietary data?
Become a referral partner from anywhere we support, get your link and introduce an owner or authorized decision-maker.
Refer a company →I own a business
Explore licensing your company's data to AI developers worldwide. Start a short assessment; no uploads needed.
Start an assessment