Synthetic training data
AI-generated data used to train AI, created because the supply of high-quality human-written text is running low.
Ranks moderate (59/100) — set apart by rising momentum (78) and recent news intensity (29).
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.
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.
Who makes Synthetic training data
The companies exposed to Synthetic training data
How to think about it
- Data is now manufactured, not just found
- Verification quality bounds synthetic-data value
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
Frequently asked
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 — 2 companies in total 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.
Go deeper on Synthetic training data
- 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
model v0.7.0 · research, not advice
Model scores are illustrative reads from our model of the AI value chain — not investment advice.