A shift in corporate spending on artificial intelligence is underway as CFOs and boards seek to curb inefficient expenditures. Traditionally, companies have relied on the most powerful AI models for all tasks, regardless of their complexity. However, with rising costs, many are now questioning this approach.
Model routing is emerging as a solution, allowing companies to match tasks with the appropriate AI model, directing complex queries to expensive frontier models while utilizing cheaper alternatives for simpler tasks. Scott Wu, CEO of Cognition, highlighted that this could lead to cost efficiencies of five to ten times for routine work.
Currently, about 95% of enterprise AI usage still relies on these costly models, even for tasks that could be handled by less expensive options. Cisco's chief product officer, Jeetu Patel, noted that the costs associated with AI can reach up to $900 million annually for large companies, prompting a reevaluation of resource allocation.
Cognition has introduced an AI productivity guarantee to address concerns about return on investment, focusing on the actual engineering hours saved rather than just activity metrics. This trend could shift pricing power from AI providers to companies, as they seek to optimize their spending.
While frontier labs will still command a premium for complex tasks, the overall market dynamics may change, influencing the valuations of leading AI companies significantly