Updated Feb. 5, 2026

Trends in AI

Frontier AI systems are advancing rapidly from increases in compute, hardware performance, software efficiency, and investment. This dashboard explores those dynamics.

Show growth values in

Model capabilities

14ECI/year

The ECI frontier has advanced by 14 points per year since the introduction of reasoning models.

Model performance
90% CI+12 to +17 ECI

New compute growth

2.5×/year
9months
0.40OOM/year

The aggregate computing power of AI chips sold each quarter is growing at a rate of 2.5×/year.

The aggregate computing power of AI chips sold each quarter is doubling every 9 months.

The aggregate computing power of AI chips sold each quarter is growing by 0.40 OOMs per year.

AI companies
90% CI2.3× to 2.7×8 to 10 months0.36 to 0.43 OOM

Training compute

5×/year
5.2months
0.7OOM/year

Training compute for frontier language models has been growing at 5× per year since 2020.

Training compute for frontier language models has been doubling every 5.2 months since 2020.

Training compute for frontier language models has been growing at 0.7 OOMs per year since 2020.

Training runs
90% CI4× to 6×4.6 to 6.0 months0.6 to 0.8 OOM

Build time

2.1years

Gigawatt-scale data centers can be built in about 2 years.

Data centers
90% CI1 to 3.6 years

Largest AI data center

1,100,000H100e

The largest known AI data center has a computing capacity equivalent to 1.1 million NVIDIA H100 chips.

Data centers

Chip performance per dollar

1.49×/year
1.7years
0.17OOM/year

AI chip performance per dollar has grown by 49% per year since 2023.

AI chip performance per dollar has doubled every 1.7 years since 2023.

AI chip performance per dollar has grown by 0.17 OOMs per year since 2023.

Hardware
90% CI1.36× to 1.66×1.4 to 2.3 years0.13 to 0.22 OOM

What drives AI progress? The story is dominated by scale. Training AI systems with more compute, power and data has consistently led to better performance. Since 2020, the compute used to train frontier language models has increased 5× per year.

This massive scale-up in training compute comes from three sources: deploying more chips in parallel, training for longer, and leveraging increasingly powerful and cost-effective AI processors. The consequences are striking. Training costs are climbing by 3.5× annually. Today’s largest AI data centers are approaching a gigawatt of power capacity — enough to power roughly a million US homes. These trends appear set to continue through 2030.

Model performance

The ECI frontier has advanced by 14 points per year since the introduction of reasoning models.

The ECI frontier has advanced by 14 points per year since the introduction of reasoning models.

Following the introduction of the first reasoning models (OpenAI’s o1-mini and o1-preview) in September 2024, the Epoch Capabilities Index (ECI) frontier has advanced linearly by 14 points per year, compared with 6 points per year for non-reasoning models.

Learn more

SOTA vs. consumer hardware models

8months

Frontier AI performance becomes accessible on consumer hardware within a year.

Model performance

LLM context windows

30×/year
2.4months
1.5OOM/year

LLM context windows have grown by 30× per year since 2023.

LLM context windows have been doubling every 2.4 months since 2023.

LLM context windows have grown by 1.5 OOMs per year since 2023.

Model performance
90% CI10× to 50×2.1 to 3.6 months1.0 to 1.7 OOM

Data on AI

Benchmarking

Track and compare frontier model performance across 40+ evaluations

AI companies

The aggregate computing power of AI chips sold each quarter is growing at a rate of 2.5×/year.

The aggregate computing power of AI chips sold each quarter is growing at a rate of 2.5×/year.

Quarterly additions of computing capacity have grown about 2.5× per year since 2024. That’s down from roughly 6× in 2023, which saw a surge in global AI training and inference capacity following the release of the initial ChatGPT and GPT-4, led by Nvidia’s new H100 GPU.

Learn more

OpenAI revenue growth

3.2×/year
7.2months
0.51OOM/year

OpenAI's annualized revenue has grown by 3.2×/year since 2024.

OpenAI's annualized revenue has been doubling every 7.2 months since 2024.

OpenAI's annualized revenue has grown by 0.51 OOMs per year since 2024.

AI companies
90% CI3.0× to 3.4×6.8 to 7.6 months0.48 to 0.53 OOM

Frontier inference compute

30%

Only 30% of OpenAI's compute spending in 2024 was used on inference.

AI companies

Frontier AI investment

$370billion

Frontier AI labs have raised more than $370 billion.

AI companies

Data on AI

Companies

Data on key organizations and labs driving AI development worldwide.

Training runs

Training compute for frontier language models has been growing at 5× per year since 2020.

Training compute for frontier language models has been growing at 5× per year since 2020.

The amount of compute used to train frontier language models has grown exponentially. Since 2020, the trend among top-5 models has grown by a factor of ~10,000.

Learn more

Training cost

3.5×/year
7months
0.5OOM/year

The cost to train frontier language models has increased by 3.5×/year since 2020.

The cost to train frontier language models has been doubling every 7 months since 2020.

The cost to train frontier language models has increased by 0.5 OOMs per year since 2020.

Training runs
90% CI2.8× to 4.4×6 to 8 months0.4 to 0.6 OOM

Report

Training compute

Training compute of frontier AI models grows by 4-5× per year.

Data on AI

Models

Database of AI systems tracked by training compute, org, and release date.

Data centers

The largest known AI data center has a computing capacity equivalent to 1.1 million NVIDIA H100 chips.

The largest known AI data center has a computing capacity equivalent to 1.1 million NVIDIA H100 chips.

SpaceXAI’s Colossus 2 in Memphis, Tennessee has an estimated computing capacity of 1.1 million H100-equivalents, making it the largest known AI data center. Meta Hyperion is expected to have over 3x more computing capacity, at 3.7M H100-equivalents by January 2028.

Learn more

Build time

2.1years

Gigawatt-scale data centers can be built in about 2 years.

Data centers
90% CI1 to 3.6 years

Cost per gigawatt

$38billion

A typical AI data center with 1 gigawatt of IT power costs roughly $38 billion in up-front capital expenditure.

Data centers

Hyperscaler compute share

71%

Five hyperscalers own over two-thirds of global AI compute.

Data centers

Data on AI

AI data centers

AI data centers tracked via satellite imagery and permit data.

Hardware

AI chip performance per dollar has grown by 49% per year since 2023.

AI chip performance per dollar has grown by 49% per year since 2023.

Since 2023, the average dollar spent on AI chips each quarter has bought about 49% more performance each year, in constant 2025 dollars. Gains have come in spurts as spending shifts to each new chip generation — price-performance was nearly flat through mid-2024, then roughly doubled as Blackwell chips grew to a majority of new spending.

Learn more

Memory bandwidth

1.28×/year
2.8years
0.11OOM/year

Memory bandwidth of GPUs has been growing by 28% per year since 2008.

Memory bandwidth of GPUs has been doubling every 2.8 years since 2008.

Memory bandwidth of GPUs has been growing by 0.11 OOMs per year since 2008.

Hardware
90% CI1.24× to 1.33×2.4 to 3.2 years0.09 to 0.12 OOM

Adoption of numerical formats

3-4years

Widespread adoption of new numeric formats took 3-4 years in past cycles.

Hardware

FLOP/s per watt

1.34×/year
2.4years
0.13OOM/year

GPUs have become 34% more energy efficient per year since 2008.

GPUs energy efficiency has doubled every 2.4 years on average since 2008.

GPUs have become 0.13 OOMs more energy efficient per year since 2008.

Hardware
90% CI1.29× to 1.39×2.1 to 2.7 years0.11 to 0.14 OOM

Report

Trends in hardware

FLOP/s performance in 47 ML hardware accelerators doubled every 2.3 years.

Data on AI

ML hardware

Key data on AI accelerators, such as GPUs and TPUs.

Trusted by leaders at OpenAI, DeepMind,
and governments worldwide

Need deeper insights? Our team offers custom research and advisory services.

Book a consultation