Theses›Datacentre & AI compute›AI data supply

AI data supply

AI data supply — the coverage-bottleneck stack (map/spine)

  • Bear
  • Closing
  • $28B by 2035

The callThe demand is real and already priced: Scale, Surge and Mercor set the entry price, so the window waits on a reset.

D1D2R1R2R3C1foragainst D1D2R1R2R3C1foragainst
Fig. 1The balance. Drivers and catalysts against restraints, each block sized by its weight; catalysts hang dashed because they have not happened yet, sized by their odds. It tips toward the restraints.
  1. D1The public internet has been read, so the next gains come from graded expert work in coding, law, medicine and finance that labs cannot make in-house at breadth.
  2. D2Labs will not buy from a supplier owned by a rival, which keeps a neutral merchant layer alive and spreads the buying.
  3. R1Entry prices already hold the demand wave: Scale at $29B, Surge above $25B, Mercor at $10B, Innodata at 70-88x forward earnings.
  4. R2Labs commoditise generic labelling and horizontal environments by splitting them across vendors or doing the work themselves.
  5. R3Past exits were poor: Appen fell from AU$4.4B to about AU$100M and Figure Eight's $300M was partly an unearned earnout.
  6. C1Mercor's IPO sets a public price for expert data supply, which either confirms the private marks or resets them.2027
Rests onthe beliefs underneath
Supported by · confidence
What would prove itdated tests, open and resolved
20272028202920302031today123456predictionswatch 20272028202920302031today123456predictionswatch
Fig. 2Each ring is a dated test: dashed while open, struck through when overdue, ticked or crossed once resolved.
  1. 1A merchant AI data vendor announces a funding round or sale at a valuation of $10B or more between 6 Nov 2026 and 31 Jan 2027.Jan 2027 · open
  2. 2Mercor's IPO and Innodata's price: a fall in either would reopen the window.Dec 2027 · open
  3. 3Neutrality remains a structural moat: independent neutral vendors keep/grow share vs lab-captured ones (Scale post-Meta), and no single vendor becomes 'the Scale AI of environments'.Dec 2028 · open
  4. 4Neutral vendors keep or grow share against lab-captured ones; a lab-owned vendor winning share would make this wrong.Dec 2028 · open
  5. 5Merchant AI data-supply revenue reaches ~$15-20B/yr by 2030 — NOT the vendor-sourced >$100B headline.Dec 2030 · open
  6. 6Merchant data supply reaches $15-20B; above $40B or below $10B and this is wrong.Dec 2030 · open
Expressionslisted, private, technologies
  • INODInnodata Inc.
    • fair
  • NBISNebius Group N.V.
    • not screened
  • APX.AXAppen
    • not screened
     Sells human-annotated and expert training data, evaluation and an annotation platform to AI labs and tech firms.
  • T.TOTELUS Digital
    • not screened
     Sells AI training data collection and annotation (ex-Lionbridge AI) to tech firms; a TELUS subsidiary, parent ticker.
  • EXLSiMerit
    • not screened
     Sells expert annotation, RLHF and model-evaluation services plus the Ango Hub platform; now part of EXL (parent ticker).
Private · 9 private companies, 7 of them in the US.

Neurolabs (Neurolaboratories Limited ) · Soda Data NV (via acquisition of Nanny ML) · Scale AI · Surge AI · Mercor · Turing · Invisible Technologies · Snorkel AI · Labelbox

Technologies
Who has already bet on itprivate rounds, from public sources

$17B raised by 7 companies, $17B of it in the last three years. Wave under way: Large rounds in the last three years; the category is funded and getting crowded.

Backed by a top-tier venture firm: Scale AI, Mercor, Snorkel AI, Labelbox. Backed by a strategic investor: Scale AI.

Reading
Show the reading