Stanford University's 2024 AI Index report revealed a dramatic increase in funding for generative AI companies in 2023, reaching $25.2 billion. This represents an almost eightfold rise compared to previous years.
Generative AI accounted for over a quarter of all private AI investments in 2023, demonstrating its growing importance and potential in the AI industry.
Key investments in 2023 included Microsoft's $10 billion deal with OpenAI, Cohere's $270 million fundraising in June, and Mistral's $415 million funding round in December.
Despite the surge in generative AI investments, corporate spending on AI decreased by 20% to $189.2 billion in 2023. This decline was attributed to a 31.2% reduction in mergers and acquisitions.
The U.S. dominated AI investments with $67.2 billion, followed by China with $7.8 billion. However, while U.S. spending increased by 22.1%, investments in China and the EU declined in 2023.
Higher AI investments in the U.S. have led to increased salaries for AI roles. For instance, a hardware engineer in the U.S. earned an average of $140,000 compared to the global average of $86,000 in 2023.
The majority of investments, $18.3 billion, went into AI infrastructure, research, and governance. Natural language processing and customer support followed with $8.1 billion.
China's semiconductor spending was close to the U.S., with China investing $630 million and the U.S. $790 million. This increase in spending follows the 2020 global hardware chip shortage.
Training costs for AI models have risen significantly. OpenAI spent around $78 million on its GPT-4 model, while Google's Gemini model required an estimated $191 million.
New AI models require more training compute. For example, Google's 2017 Transformer model needed 7,400 petaFLOPs, whereas the Gemini Ultra required 50 billion petaFLOPs.
The report highlighted increased industrial dominance in leading AI models. Google released 40 model foundation models since 2019, while OpenAI released 20. Tsinghua University from China was the top non-Western outlet with seven releases.
A growing trend in AI is the development of multimodal AI models that can process images, videos, and text. Stanford's report emphasized the combination of large language models with robotics or autonomous agents, marking progress towards robots operating more effectively in the real world.
The Stanford AI Index report highlights the rapid growth of generative AI investments in 2023, despite a decline in overall corporate AI spending. With rising model training costs and increasing compute requirements, the AI landscape is evolving, emphasizing the importance of innovation and investment in AI technologies.
