Frontier Signal · 28 October 2026 archive

AI demand, this cut.

Frontier Signal tracks public AI data-buyer programmes and frontier-lab hiring as a map of demand. This cut: two new code buyers joined the board, one quoting $100 per million tokens of private source code, and buyers now list deal ranges by kind of company — law firms top out at $1.25M.

Latest brief Read the investigation Open watchlist

Archived public cut. Current answers live on the latest brief.

Demand mix

What is being bought.

Hot categories by mentions across verified buyer programmes this cut.

Mentions

Verified buyer programmes

Software / code — 12
Finance & accounting — 8
Legal — 8
Healthcare / medical — 6
Manufacturing — 4
Logistics — 4
Construction — 3
Insurance — 3
Agent trajectories — 2

Read

Code gets a unit price

The AIData.Broker board grew from 42 to 44 buyers (33 with a public figure). Pangea.ai quotes $100 per million tokens of private source code and $40 per million tokens of git history, and reports $6M paid to code owners in 60 days across 11,069 repositories. Idler asks companies with 10 to 100 employees to license their codebase or de-identified business data. Both figures are as published on the board.

Deal ranges by kind of company

What buyers list, per kind.

Ranges from buyers’ own listings, as collected on the AIData.Broker board this cut.

Professional

Law firms lead

Law firms $50,000 to $1,250,000
Accounting and tax firms $25,000 to $865,000
Investment firms $50,000 to $800,000
Cybersecurity companies $20,000 to $750,000

Health & science

Four health kinds now priced

Digital health companies $100,000 to $900,000
Biotech and life sciences labs $50,000 to $875,000
Medical practices and clinics $80,000 to $820,000
Imaging and diagnostics centres $80,000 to $620,000

Media

Authored media near $1M

Game studios $100,000 to $1,000,000
Video studios $95,000 to $870,000
Publishers and education companies $35,000 to $370,000

Industrial

Materials on the list

Semiconductor and materials companies $100,000 to $750,000.

Bounty / intensity

Scarcity and quality — not one number.

Referral bands and “up to” rates are signals. Do not collapse them into a rate card.

micro1

$100K–$500K / $500K–$2M / $2M+

Licenses qualifying agent trajectories for $1–2M a year. Pays in full within 30 days of acceptance.

Operating history

Lazarus up to $1M

Lazarus (Turing): payout up to $1M for company operating history; Refer & Earn up to $10k.

Expert seats

Mercor and Handshake AI

Mercor referrals $250–$15,000. Handshake AI expert rates up to $125/hr.

Intro markets

Mode / BizBuyData staff floors

Its own site takes legal from 6+ staff, accounting, finance and medical from 10+, others from 30+; its referral programme pays on 10+, 15+ and 50+ (per the AIData board). Referral figures $5K–$50K+.

Lab hiring as roadmap

Verified counts this cut.

Human Data and data-ops seats, not keynote language. Titles unchanged since 27 October.

Anthropic

681 Greenhouse jobs

679 on 27 October. Data Operations Manager, Human Data; Product Management, Human Data Platform.

xAI

305 Greenhouse jobs

308 on 27 October. Human Data — Engineer; Human Data Manager.

OpenAI

805 Ashby postings

796 on 27 October. Same six Human Data / data-acquisition roles.

Walls

Consent, de-id, privilege

OpenAI wants no sensitive personal data; Lazarus and Idler ask for de-identified records; legal and finance files carry privilege. HUMXN and Grepped stay parked; Ooak unreachable from our box two days running.

Cross-signal

Where the two streams agree.

Code demand now shows up on both sides: two new code buyers on the board, Lazarus and micro1 already buying repos and operating history, and Anthropic, xAI and OpenAI still staffing Human Data.

Idler’s own site sells reinforcement-learning environments built from real production work — the same direction as micro1 and SuperAnnotate pricing agent trajectories.

What to open next

The spine is the frame. The watchlist is the list.

This page is the archived public cut for 28 October 2026. Use the investigation page for the questions. Open the gated watchlist for the full answers.

Latest brief Read the investigation Open watchlist