No matter where you look or where you go, it’s impossible to escape the trend of artificial intelligence. Although this technology has coexisted with us for decades, new tools like ChatGPT have played a crucial role in popularizing AI among the general public.
Despite the interest and fascination it generates in us, the industry faces a significant challenge: how can this technology be monetized?
To begin with, getting AI-powered generative models up and running has proven to be quite expensive. High energy consumption and the need for powerful servers make the cost of using this technology very steep. Because of this, companies like Google or Adobe are still experimenting with different strategies to develop and sell this technology.
Success doesn’t equate to profits
For instance, we have learned that Microsoft has incurred significant losses due to GitHub Copilot, one of the first generative AIs they launched in collaboration with OpenAI. This service assists programmers in developing, correcting, and translating code, making it extremely popular (in fact, more than 1.5 million people have already used it). However, its success doesn’t offset the high operational costs.
In the case of GitHub Copilot, users pay around $10 per month to use the assistant. Nevertheless, in the early months of 2023, Microsoft was losing an average of over $20 per user (some even as much as $80).
To prevent this, Adobe and other companies are limiting its use and charging users based on their usage. As expected, this is not well-received by the public.
According to Adam Selipsky, CEO of Amazon Web Services, “many customers […] are unhappy about the high price they have to pay to use some of these models.” As it is a relatively new technology, the public still needs to understand what kind of use they want to make of artificial intelligence and, above all, at what price. Unlike other types of software, generative models require more money and years of development to succeed.
For businesses, it’s just a matter of time before the production of generative models becomes more affordable, similar to what has happened with other technologies.
