A recent analysis shows that over 70% of cloud environments are using managed AI services, with OpenAI being a key tool, either directly or through Azure SDKs. This suggests that generative AI tools are rapidly becoming standard in cloud business models.
The study, conducted by cloud security platform Wiz, looked at more than 150,000 public cloud accounts and found that Microsoft is leading the way. Around 70% of Azure environments analyzed included Azure AI Service instances, which accounts for about 39% of all cloud environments. The report also highlighted that the use of Azure OpenAI grew by 228% over a four-month period in 2023.
While Azure is ahead, this may not be surprising, as AWS’s fully managed service, Amazon Bedrock, only became generally available in September. However, Amazon’s SageMaker is just behind Azure AI Services, with a 38% deployment rate. Bedrock, though released during the research period, wasn’t included in the analysis, but preliminary data suggests that at least 15% of organizations are already using it.
Most users of managed AI services are still in the experimentation phase, according to the study. About 32% of users fall into this category, while 28% are considered active users, and 10% are power users.
Wiz based its findings on the number of service instances in each cloud environment, though it wasn’t able to define instances by workload type (e.g., development or production). Power users are defined as those with 50 or more instances, which might seem low but is understandable given the high cost of training and fine-tuning AI models. Additionally, some providers have strict limits on how many AI service instances a customer can deploy.
The study also revealed that 53% of cloud environments are using OpenAI or Azure OpenAI SDK, which allows integration with OpenAI models like GPT and DALL-E. It also noted the high prevalence of self-hosted AI and ML software in the cloud, with 45% of environments using Hugging Face Transformers, 32% using LangChain, and 22% using TensorFlow Hub.
Looking ahead, Wiz pointed out that the cost of training and inference will be a critical focus for organizations over the next year. This will create a turning point as companies start to assess the true potential of AI technology.
The report concludes that 2024 will likely be a pivotal year, with many organizations deciding which AI paths are worth pursuing. As more companies experiment with generative AI, it will become clear whether this technology can drive efficiency and create new, groundbreaking features.
