Funding for generative AI surges

2 mins read

A new report from Stanford University shows that funding in the generative AI space surged dramatically in 2023, driven by the likes of OpenAI and Anthropic who both recorded substantial increases in capital.

Credit: putilov_denis

Stanford’s 2024 AI Index revealed a nearly eightfold increase in funding for generative AI firms, reaching $25.2 billion in 2023. In fact, generative AI now accounts for more than one-quarter of all AI-related private investment made last year.

Among the most significant investments made last year were Microsoft’s $10 billion OpenAI deal and Cohere’s $270 million raised in June 2023.

Stanford’s report did show, however, that corporate spending on AI dropped in 2023, decreasing by 20% to $189.2 billion and this was attributed to a reduction in mergers and acquisitions which fell 31.2% from the previous year. Despite the drop, nearly 80% of earnings calls for Fortune 500 firms mentioned AI.

Investments were dominated by firms from the US, with $67.2 billion invested, almost nine times greater than the amount invested by the second-highest spender, China with $7.8 billion.

The report found that private investment in AI dropped in China and the EU in 2023 compared to 2022, while US spending rose by 22.1%.

Those areas that attracted most investment in 2023 included AI infrastructure, research and governance with a total of $18.3 billion. Stanford’s report stated that this spending reflected big players including OpenAI and Anthropic building large-scale applications like GPT-4 Turbo and Claude 3.

The second highest spending sector was natural language processing and customer support $8.1 billion.

The US was the largest spender across all the various AI technology categories apart from facial recognition where China spent $130 million compared to $90 million in the US.

In terms of semiconductor spending, China spent $630 million compared to the $790 million spent by the US. That investment comes as governments around the world have been increasing semiconductor spending to shore up supply chains following the 2020 global hardware chip shortage.

With companies like OpenAI raising millions of dollars in funding, Stanford’s report notes that those same firms are raking up big bills from training new models.

Model training costs rose in 2023, with Stanford’s researchers suggesting investments in large-scale foundation systems were a leading factor behind the rise.

Stanford’s AI Index reported that the training costs for advanced AI models have risen significantly. For example, OpenAI spent an estimated $78 million to train its GPT-4 model, while Google's flagship Gemini model required an estimated $191 million - earlier models were much less expensive.

Stanford collaborated with Epoch AI to come up with estimates on training costs and those figures were based on information from related technical documents and press releases.

In addition to costing millions to train, AI models trained in the past year used more training compute. Google’s Gemini Ultra, for example, required 50 billion petaFLOPs.

Stanford warned that power-intensive systems like Gemini Ultra were becoming increasingly inaccessible to academia due to their sheer costs involved to run them.

Google was found to be releasing the model foundation models, publishing 40 since 2019. OpenAI stands in second place with 20. The top non-Western outlet releasing AI models was China’s Tsinghua University with seven.

The majority of large-scale AI systems published in 2023 stemmed from the US, with 109. Chinese institutions were second, but only managed 20.

One growing trend highlighted in the report was the growing number of multimodal AI models or systems that can process images or videos as well as text.

"This year, we see more models able to perform across domains," said Vanessa Parli, Stanford HAI director of research programs. "Models can take in text and generate audio or take in an image and generate a description. An edge of the AI research I find most exciting is combining these large language models with robotics or autonomous agents, marking a significant step in robots working more effectively in the real world."