Second report of a series? from a survey of enterprise AI buyers. OMG this stuff gets expensive fast and you can’t throttle access to it because users will go around you.
Large enterprise
In this report, we surveyed a 1000 more enterprises on the hidden costs and challenges of Generative AI business adoption. We show how AI leaders are navigating the uncharted territory of Gen AI and running into the unfamiliar and unpredictable. We discovered:
- Companies are badly underestimating the TCO of Generative AIMost companies want to implement and run Gen AI themselves
- Orgs are experimenting with a huge array of models and tools
We focused on C-suite and team leaders with job titles like CIO, CTO, Head of AI, VP of Data or VP of AI, across a range of verticals.
Get it now. FREE.
Name *
Email *
Key Charts
We tracked the cost drivers of Gen AI and found they range across the board, everything from GPUs, to highly specialized talent, to retraining people.
The costs skyrocket when it comes to training custom models with everything from data prep, along with tools and infrastructure hitting the bottom line.
But we also found that many companies are dramatically underestimating costs for running AI in terms of inference on-prem or in the cloud and they’re really underestimating token based or subscription pricing for cloud models.
We also looked at how companies use Gen AI. Many are training their own models in-house but most are using open models or fine tuning proprietary models, rather than bear the cost of building a base model from scratch, which is time consuming and expensive from a people and compute perspective.
People are also bypassing the whole training and fine tuning process altogether by using complete AI driven apps, a trend we expect to accelerate in the coming years.
Many organizations are looking to staff in-house to support these apps and that means we’re going to have a big talent crunch in the coming years, as companies compete to hire the best and brightest.
It seems enterprises are moving quickly to generative AI but they may not be understanding the problems and challenges with it well enough. They're going to have to learn fast to compete at peak performance.
Fortune 500 CTO
