are planning to create their own enterprise-specific LLMs 51% Q. Are you planning to create your own enterprise-specific LLMs for use in Generative AI implementations? Although these powerful AI models can bring strategic value to business opera�ons, the cost implica�ons can be prohibi�ve, with the development and maintenance of LLMs poten�ally reaching into the millions of dollars. Many companies are looking at smaller language models as workarounds, but these require exper�se to customize. Nevertheless, significant financial outlays were not iden�fied as a top concern in our survey—“cost of deployment” for AI implementa�ons ranked 7 th out of a possible 10 challenges—but it presents a substan�al considera�on for companies, especially when viewed alongside the extensive computa�onal resources and infrastructure that crea�ng an enterprise-specific LLM would likely demand. But without adequate KPIs for AI-enabled opera�ons, proving ROI is challenging. As companies plan their strategy for AI technology investments, the ROI must be eviden�al via key performance indicators or they risk losing budget. However, our survey reveals that 72% of business execu�ves struggle to measure the success of their AI implementa�ons effec�vely, making it challenging to secure funding for more advanced AI projects. (Another 8% are not even aware of any useful KPIs they can apply to AI-enabled opera�ons.) Yet given the fast-paced evolu�on of the AI landscape over the past couple of years, this lack of adequate assessment metrics is not par�cularly surprising. Large language models basically repackage the internet and that repackaging is useful and helpful in many places. But if the thing that you're interested in automa�ng or doing is not embedded somewhere on the internet, then it's not going to happen. The harder task will be to assemble a package of data that knows the things that are of interest in the business process. – Peter Reinhardt, CEO & Co-founder, Charm Industrial TCS AI for Business Study Key Findings Report 25