Skip to content

Models layerTraining

Synthetic training data

What it is

AI-generated data used to train AI, created because the supply of high-quality human-written text is running low.

2companies exposed3parts need it
Where Synthetic training data sitsLayer 4 of 5
EnergyChipsInfraModelsAppsSynthetic training data
Model relevance54/100

Ranks moderate (54/100) - set apart by rising momentum (78) and fragility to shocks (62).

The short read

Why Synthetic training data matters, and what breaks without it.

Why it matters

It is becoming the primary way to keep scaling training and to manufacture the verifiable data that reasoning RL requires.

Why now

Credible projections put exhaustion of high-quality human text in the late 2020s, so frontier labs now lean heavily on synthetic and self-generated data.

If Synthetic training data runs short

Low-quality or unfiltered synthetic data degrades models (model collapse), quietly capping capability gains.

Exposure

The companies exposed to Synthetic training data

2 companies are mapped to Synthetic training data, ranked by how strongly each one is exposed. The same ranking as the full Synthetic training data list.

Who makes Synthetic training data

Tagged as producer or supplier: NVIDIA and Scale AI.

The Synthetic training data ranking

Editorial read

In depth · editorial + model

Model-generated corpora, self-play traces and verifiable problem sets used to augment or replace scarce human data in training and post-training. It is becoming the primary way to keep scaling training and to manufacture the verifiable data that reasoning RL requires. Credible projections put exhaustion of high-quality human text in the late 2020s, so frontier labs now lean heavily on synthetic and self-generated data. Low-quality or unfiltered synthetic data degrades models (model collapse), quietly capping capability gains.

Mental models

How to think about it

  1. 01Data is now manufactured, not just found
  2. 02Verification quality bounds synthetic-data value

Signals to follow

What to watch

  • Epoch-style estimates of human-data exhaustion
  • Share of training tokens that are synthetic
  • Evidence of model-collapse vs quality-filtered gains

Questions

Frequently asked

The short answers about Synthetic training data, from the same model as the rest of the page.

What is Synthetic training data?

AI-generated data used to train AI, created because the supply of high-quality human-written text is running low.

Why does Synthetic training data matter for AI?

It is becoming the primary way to keep scaling training and to manufacture the verifiable data that reasoning RL requires.

Who makes Synthetic training data?

The companies the model tags as producers or suppliers of Synthetic training data: NVIDIA, Scale AI.

Which companies are most exposed to Synthetic training data?

NVIDIA, Scale AI. In total, 2 companies are mapped to Synthetic training data.

What happens if Synthetic training data runs short?

Low-quality or unfiltered synthetic data degrades models (model collapse), quietly capping capability gains.

Where does Synthetic training data sit in the AI value chain?

Synthetic training data sits in the Models layer of the AI value chain.

In the app

Go deeper on Synthetic training data

Inside THE ENTITY, Synthetic training data is live: its pressure, its links and the companies on it, refreshed nightly.

  • The materials, geographies and policies it depends on, heat-mapped
  • Substitutes, relief valves and the domino chains if it tightens
  • The live tension score, momentum and news drivers
  • Four levels of analysis, from plain English to strategic
The whole chain

Research, not investment advice. Model scores are illustrative reads from our model of the AI value chain, and companies are mapped to explain the chain, never as recommendations.

as of 2026-11-09Medium confidence model v2.01