In recent years, the artificial intelligence industry has developed rapidly, and large-scale language models have emerged one after another. However, the editor of Downcodes learned from Cohere CEO Aiden Gomez’s recent podcast that competition in the AI industry is fierce and the business of selling AI models is heading towards “zero profit.” The high cost of model training and fierce price competition have put many companies under tremendous pressure to survive. This article will provide an in-depth analysis of the current difficulties and future development trends of the AI industry.
In a recent podcast, Cohere CEO Aidan Gomez said that competition in the current AI industry is very fierce, and the business of selling AI models is quickly becoming a "zero-profit industry." He mentioned that companies like OpenAI and Anthropic invest billions of dollars every year in training large models like GPT-4 and Claude, but due to fierce price competition, the overall business has become extremely fragile.
Picture source note: The picture is generated by AI, and the picture is authorized by the service provider Midjourney
Gomez pointed out that the current price war for AI models has almost eliminated the industry’s marginal profits. "If you only rely on the sales model, the next days will be very difficult." He said that the "sales model" here refers to the service that provides developers with API access to these AI models, and in this field, OpenAI, Companies like Anthropic, Google, and Cohere are all facing similar dilemmas.
AI model R&D companies are facing the challenge of continuously improving model performance while also paying large fees to hardware manufacturers such as Nvidia. At the same time, various companies have lowered the access price of their models in order to retain users, and the open source models of companies such as Meta are free for people to use.
Gomez also mentioned that although the current AI model is facing huge losses, he believes that Cohere's AI model will be an attractive business in the long term. In order to generate revenue in the short term, launching related products is also a feasible path. He emphasized that the current AI model business is actually operating at a loss.
For giants like Microsoft and Google, they can absorb these losses through financial subsidies, but for startups, this is not the case. Cohere is one of the few startups still developing cutting-edge AI models, alongside OpenAI, Anthropic and Mistral. Many similar startups, such as Inflection and Character.ai, have been acquired by large cloud service providers, making their businesses unprofitable while retaining their technology.
Gomez also mentioned that relying on cloud service providers is not a good business model. He pointed out that venture capitalists only care about returns, while cloud service providers may have other motives. "This is very dangerous for startups."
Although the industry is full of expectations for the innovation of AI models, it is still unknown when profitability will be achieved. Obviously, not every AI startup can wait until that moment.
Highlight:
Intensified competition has made the sales profit of AI models almost zero, and companies need to face the pressure of price wars.
? Start-ups find it difficult to survive under the subsidies and low-price strategies of giant companies and face huge pressure to survive.
Although the current business model suffers serious losses, in the long run, the market still has huge demand potential for AI models.
All in all, competition in the field of AI models has entered a fierce stage. How to survive in the fierce price war and find a profit model is a huge challenge facing many AI companies. In the future, the profit model of the AI industry may need to be innovative in order to occupy a favorable position in long-term competition. The editor of Downcodes will continue to pay attention to the dynamics of the AI industry and bring more exciting reports to readers.