Author: IRPA AI Analyst & Former White House Transformation Advisor, Chris Surdak
OpenAI recently announced new pricing for their most advanced models, and they point towards a potentially rocky road ahead for adoption of their solutions. In an anonymous release by company sources, OpenAI announced pricing of AI agents ranging from $2,000-$20,000 per month. With such pricing, will these tools still provide a value proposition beyond human workers, or is this potentially pricing AI out of the market?
Sources stated that the upper-end of these agents will provide PhD-level skills and knowledge, and could be applied to a wide range of advanced topics and processes. But, as another layer of labor arbitrage one must wonder if it is possible to keep such agents engaged enough and productive enough to justify their high cost. While the output of these tools has been improving, there is still a distinct “AI style” to their outputs. It’s often easy to tell when you’re reading AI-generated material, and one wonders if this would be acceptable coming from an agent which costs a quarter million dollars per year.
OpenAI is under extreme pressure to generate revenues, let alone profits. Their burn rate is clearly in the many billions of dollars per year, even with their current pricing of $200 per month for their advanced model. How many viable customers would exist at these even higher price points, and how less effective will the lesser price point model become are open questions, but it is clear that OpenAI is struggling to operate as a viable business, while facing ever-increasing competition from other platforms.
Given that each week seems to present another new release of yet another new model which provides yet more evidence of improving performance demonstrates that none of the hyperscalers are able to create meaningful competitive advantage. This inability to create a “moat,” or sustainable strategic advantage is a significant issue in this space. Customers may be extremely reluctant to commit to an extremely expensive agent whose performance may be eclipsed the following week by a competitive product.
These uber-agents may also exacerbate the oversight problem. If the agents are producing PhD-level outputs they will require equally-or-better-qualified humans to review and approve their work. The instances of generative AI hallucinating or creating completely made-up responses are well known, hence their outputs nearly always need to be validated before their put to use. Renting a fleet of uber-agents means having an equally-talented team of humans to provide oversight. As such, one wonders if these agents will actually generate a positive return on investment, or will organizations get better returns by simply retaining their best human workers.
About the Author: Chris Surdak
Chris Surdak is a Senior IRPA AI Advisor and was formerly White House Chief Transformation officer, Automation & AI Practice Lead at EY & Executive Partner for Digital Transformation at Gartner. He’s an engineer, futurist, transformation executive and best-selling author, with over 30 years’ experience in technology development and deployment, digital transformation, blockchain, data and analytics and AI & intelligent automation.
CLICK HERE TO SCHEDULE AN ANALYST CHAT
Links:
Originally posted on 2025-03-07 in the IRPA AI Network — Enterprise AI