The "2025 First Quarter AI Application Report" mainly analyzes how major global enterprises used large models in the first half of 2025, and what is different this year compared with last year. The report points out that about 45% of organizations have already used AI in production environments, with engineering and R&D being the leading areas of AI adoption. AI chat applications (such asChatGPT) and coding tools (such asGitHub Copilot) are widely popular. In terms of language models,Google Geminiand OpenAI's GPT series occupy a dominant position. Among multimodal AI models (voice, image, video), OpenAI performs prominently. The inference services market is dominated by first-party APIs (such as OpenAI, Google). NVIDIA occupies a dominant position in the AI training hardware market. The report discusses the challenges of AI adoption, including intelligence, reliability, and cost issues.
Get the "2025 First Quarter AI Application Report" PDF original file, scan the QR code to follow and reply: 20250812
AI Adoption Overview
- AI Adoption Maturity
- AI Applications in Production Environments: The report points out that about 45% of organizations have already used AI in production environments.
- AI Adoption Stages: In addition to production environments, 23% of organizations are in the prototype development stage, 27% are in the exploration stage, and 5% have not yet started.
- Industry and Regional Distribution of AI Adoption
- Industry Distribution: The report covers multiple industries, including technology, education, government, nonprofit organizations, and others. Among them, the technology industry is the most active in adopting AI.
- Regional Distribution: From a regional perspective, the United States and Europe have relatively high AI adoption rates, while emerging markets such as China and India are also rapidly catching up, demonstrating the broad global influence of AI technology.
Key Trends in AI Adoption
- AI in Production: About 45% of organizations have already used AI in production environments, showing the shift of AI technology from prototype development to actual production environments.
- Diversification of AI Use Cases: Organizations are applying AI in multiple fields, with engineering and R&D being the most important field, followed by customer support and sales and marketing.
- Choice of AI Models: Google Gemini and OpenAI's GPT series are the most popular AI models, and DeepSeek is the preferred open-weight model.
- Attitudes Toward Chinese AI Models: If hosted on infrastructure outside China, 55% of respondents are willing to use models from Chinese AI labs.
AI Chat Applications and Coding Tools
- AI Chat Applications: ChatGPT is currently the most popular AI chat application, followed by Gemini and Claude.
- AI Coding Tools: GitHub Copilot and Cursor are the most popular AI coding tools, ahead of other tools such as Claude Code and Gemini Code Assist.
Language Models (LLMs)
- LLM Family Preferences: Google Gemini and OpenAI's GPT series are the most popular LLM families, and DeepSeek is the preferred open-weight model.
- LLM Market Dynamics: Compared with 2024, the number of LLM uses and considerations increased significantly in 2025, indicating improved market maturity and demand for experimentation.
Multimodal AI Models
- Voice Generation: OpenAI and ElevenLabs are the most popular speech generation models, and streaming quality, natural speech quality, and latency are the most important factors when choosing a model.
- Image Generation: OpenAI leads in image generation models, and what users value most is prompt adherence.
- Video Generation: OpenAI and Google lead in video generation, and what users value most is prompt adherence and realism.
Inference Services
- Inference Service Providers: First-party APIs (such as OpenAI, Google, and Anthropic) and chip challengers (such as Groq and Cerebras) dominate the inference services market, while Amazon and Azure have seen a decline in market share.
Training and Hardware
- Training Accelerators: NVIDIA dominates the AI training accelerator market, with about 78% of respondents using NVIDIA accelerators, while the use of Google and AMD accelerators is lower.
Report Conclusion
The report points out that AI technology is rapidly moving from the laboratory to practical applications, and organizations face challenges such as intelligence, reliability, and cost when adopting AI. At the same time, market demand for AI models and tools is also constantly changing, especially in the fields of language models and multimodal generation.
Get the PDF original file of "Q1 2025 AI Application Report", scan the code to follow and reply: 20250812