Jun 18, 2025

Share Kong Inc., a leading developer of cloud API technologies, today released findings from their latest research report, What’s Next for Generative AI in the Enterprise, which highlights the insights, preferences and concerns about enterprise Large Language Model (LLM) adoption. AI is reshaping business and the global economy. Yet not enough is known about how business investments in generative AI are taking shape. The research findings paint a clear picture: LLM usage is surging, with 72% of enterprises expecting to increase spending on LLMs over the next year, with nearly 40% of respondents indicating that their investment would exceed $250K this calendar year. “AI adoption is accelerating at an unprecedented rate, and organizations that can keep up will have a clear competitive edge,” said Marco Palladino, CTO and Co-Founder of Kong Inc. “This report highlights that while enterprises are moving in the right direction, future success will be dependent on having a robust AI infrastructure in place to help remove common adoption roadblocks and provide a secure, scalable foundation, allowing organizations to capitalize on new GenAI opportunities.” In 2024, Kong’s survey data showed that OpenAI’s models were the most widely used at 27% with Google Gemini being used by 17% of respondents. However, in the first quarter of 2025, Google saw a rapid rise in usage, with 69% of respondents reporting using Google’s models in the last 90 days; 55% say they used OpenAI models in the same timeframe. Meta and IBM models trail at 38% and 26% respectively. These findings demonstrate that Google is making significant strides with its AI capabilities within the tech community. As background, IT professionals and developers are often using generative AI to automate coding tasks, generate documentation, test APIs, and accelerate software development workflows. This study also highlights important insights into the use of one of the newest LLM entrants, DeepSeek, which has garnered both attention and concerns due to potential foreign influence. Despite these concerns, DeepSeek has already gained significant popularity with 80% of respondents saying they use or would consider using it. However, of those not using it, 68% cited privacy concerns as the deterrent, with 46% citing internal mandates prohibiting its use. Related Posts Challenges and Opportunities with LLM Adoption In general, data privacy and security concerns are the most persistent challenges enterprises face to AI adoption at scale, according to 44% of respondents. Second to that are cost and budget constraints (24%), as deploying LLMs requires significant investment in computing infrastructure, cloud resources and ongoing fine-tuning to ensure relevance. Additionally, 14% of respondents cited integrating into existing systems as a top challenge to adoption, highlighting the technical complexity of embedding AI into legacy architectures. While developer sentiment around GenAI has been mixed over the past two years, 46% of developers view it as a tool for enhancing productivity and innovation with companies increasingly leveraging GenAI to automate repetitive tasks, accelerate content creation, assist with data analysis, and enhance decision-making. Additional findings from the report include: 31% of respondents named security and data privacy compliance as their top factor when selecting an LLM provider. 63% use paid or enterprise versions of LLMs, with only 17% relying on free tiers. AI-powered chatbots (27%) and code generation & developer productivity (26%) have emerged as the dominant LLM use cases. 51% say open source LLMs are or will be superior to proprietary models; 37% prefer a hybrid approach. Microsoft Azure AI (45%) was cited as the platform most used within the last 90 days to serve/consume a model with OpenAI Platform (41%) and Google Vertex AI (35%) following close behind. This report examines the rate of adoption of large language models, including the opportunities and limitations for businesses. To gain insights, a survey was commissioned of a representative sample of 550 IT leaders, software developers, engineers, and Kong users. Responses were gathered from February 26 to March 31, 2025. [To share your insights with us, please write to  psen@itechseries.com]