This chart illustrates the exponential growth in training compute required for artificial intelligence models between 2020 and 2025, emphasizing the escalating resource demands of modern AI development. It compares three major sectors, including large-scale commercial models, industry research and development, and academic or open-source projects, showing how computational requirements diverge as model complexity increases. The sharply rising curve for large-scale models demonstrates how cutting-edge AI systems depend heavily on massive processing power, while industry and academic efforts follow steady but distinct growth paths. This visual is well suited for technology strategy reports, AI infrastructure planning, educational materials, and business presentations that focus on the cost, scale, and future trajectory of artificial intelligence research.
| Year | Large-Scale Models | Industry R&D | Academic & Open-Source |
|---|---|---|---|
| 2020 | 120 | 80 | 40 |
| 2021 | 350 | 220 | 110 |
| 2022 | 900 | 520 | 260 |
| 2023 | 2,800 | 1,400 | 520 |
| 2024 | 8,500 | 3,200 | 900 |
| 2025 | 22,000 | 6,800 | 1,500 |
