With the rise of artificial intelligence agents, businesses are facing a surge in costs related to this technology. To manage these increasing expenses, many organizations are now looking for more economical solutions, turning to AI models that are less demanding in resources. This article explores the strategic decisions they are making to utilize AI while adhering to their budgets.
An Inevitable Rise in Costs
The use of AI agents has seen exponential growth, resulting in skyrocketing bills for many companies. This phenomenon has been exacerbated by the growing popularity of generative models such as ChatGPT, which were initially offered at very attractive prices by industry giants like OpenAI. However, this period of “subsidized intelligence” seems to be over, and companies must quickly find solutions to address the rising costs associated with the use of these technologies.
Pricing Adjustments by AI Giants
Major players in artificial intelligence are starting to revise their pricing based on the actual use of their computing resources. The increase in demand for more complex tasks not only leads to higher infrastructure costs but also a multiplication of tokens, the unit of measurement used to evaluate the work performed by AI. For a given task, new configurations may indeed require resources that far exceed those needed for a simple request to a chatbot.
Businesses Facing an Economic Dilemma
In this context of rising costs, several companies, including some major brands like Target and Starbucks, are starting to question the necessity of adopting expensive AI solutions at all costs. According to experts, in some cases, the expenses related to AI can even surpass those of a human employee in the short term. As a result, companies must be cautious and judicious in their choice of integrating advanced technologies.
An Alternative Towards Less Costly AI
To meet the growing demand for more affordable artificial intelligence, many businesses are turning to less powerful yet also less expensive models, such as open weight models and small language models (SLM). These latter solutions, which can often be run on local servers or computers, offer businesses the opportunity to significantly reduce their costs of accessing cloud services, avoiding excessive usage fees to the provider.
Optimizing AI Usage with Discernment
In the quest for a more economical use of AI, some businesses are also seeking to optimize their processes by breaking unique requests into clear steps. By assigning each step to the most suitable agent, companies can lower their usage costs. For example, a monolithic model may cost up to $15 for a million tokens, while smaller models could bring this cost down to just 5 cents, a significant difference that can impact the overall budget of companies.
A Market in Full Transformation
The transition to more accessible models is opening a new market, where platforms for selecting and coordinating AI agents are rapidly developing. Innovative startups are seeking to compete with established companies like Amazon, which already offers a variety of solutions using its Bedrock platform. The choices available range from using highly specialized models to economical options, such as that of Anthropic with its Haiku model.
As the technological landscape evolves, the ability of companies to adapt to these new solutions will determine their success in a sector where efficiency takes precedence over raw performance. The demand for smart technologies remains strong, and more advanced users will continue to invest in the best, predicting a promising future for artificial intelligence in the modern economy.
