Will frontier AI effective training compute increase by a factor 10 billion between 2025 and 2035?
12
100Ṁ4902035
61%
chance
1H
6H
1D
1W
1M
ALL
Resolves as YES if the effective compute involved in the training of frontier AI models increases by a factor of a least 10 billion between January 1st 2025 and January 1st 2035.
In the context of this question, we follow the definitions proposed by Epoch AI, where effective compute is defined as the composite of algorithmic progress and compute scaling ("algorithmic progress decreases the pretraining cost of AI").
Links:
https://x.com/EpochAIResearch/status/1890530253866422525/photo/1
This question is managed and resolved by Manifold.
Market context
Get 
1,000 to start trading!
1,000People are also trading
Will an AI achieve >85% performance on the FrontierMath benchmark before 2027?
35% chance
Efficient full homomorphic encryption for frontier model training by 2030?
11% chance
Most training run compute greater than 2e27 FLOP by EOY 2026?
92% chance
Will compute get a million times more energy efficient by 2075?
11% chance
Will an AI achieve >80% performance on the FrontierMath benchmark before 2027?
50% chance
AI model training time decreases fourfold by mid-2027?
36% chance
Will an AI achieve >85% performance on the FrontierMath benchmark before 2028?
64% chance
Before 2028, will there be enough inference capacity to generate 30T frontier model tokens per day?
99% chance
100GW AI training run before 2031?
33% chance
AI: Will someone train a $10T model by 2100?
59% chance
People are also trading
Related questions
Will an AI achieve >85% performance on the FrontierMath benchmark before 2027?
35% chance
Efficient full homomorphic encryption for frontier model training by 2030?
11% chance
Most training run compute greater than 2e27 FLOP by EOY 2026?
92% chance
Will compute get a million times more energy efficient by 2075?
11% chance
Will an AI achieve >80% performance on the FrontierMath benchmark before 2027?
50% chance
AI model training time decreases fourfold by mid-2027?
36% chance
Will an AI achieve >85% performance on the FrontierMath benchmark before 2028?
64% chance
Before 2028, will there be enough inference capacity to generate 30T frontier model tokens per day?
99% chance
100GW AI training run before 2031?
33% chance
AI: Will someone train a $10T model by 2100?
59% chance