AI Adoption by SMEs – Key Insights (OECD, G7 2025)
1. The core problem: SMEs are falling behind on AI
Across G7 economies, small and medium-sized enterprises adopt AI far less than large firms, even though AI is becoming easier to use and more accessible. While AI use is growing, it still lags far behind other digital technologies like cloud computing. This gap is persistent across countries, sectors, and types of AI applications
OECD Discussion Paper
Importantly, SMEs that do use AI often apply it to peripheral tasks, such as marketing or drafting content, rather than embedding it into core business operations like production, logistics, or R&D.
2. Why this matters: productivity and competitiveness
Evidence reviewed by the OECD shows a strong association between AI use and higher productivity at the firm level. Firms using AI tend to be more productive than similar firms that do not—though part of this advantage reflects stronger digital capabilities overall.
OECD Discussion Paper
3. The OECD taxonomy: not all SME AI users are the same
To avoid “one-size-fits-all” policies, the OECD proposes a taxonomy of SME AI adopters based on:
• Digital maturity
• Complexity of AI use
• Scope of AI application
This helps explain how SMEs adopt AI and what kind of support they need.
4. Real-world examples of the OECD AI adoption taxonomy
AI Novices
Profile:
SMEs experimenting with AI in limited, low-risk ways, usually using off-the-shelf tools with little integration.
Real-world example:
• A small coffee roaster in the United States uses ChatGPT to write product descriptions, improve SEO, draft marketing emails, and analyse shipping costs. AI improves time efficiency, but outputs are always manually checked.
Typical use:
• Writing
• Marketing
• Basic analytics
AI supports work but does not reshape the business.
AI Optimisers
Profile:
SMEs that integrate multiple AI tools across departments, achieving visible efficiency gains, though without building their own models.
Real-world examples:
• A French artisan bakery uses ChatGPT for customer communication and content creation, and Midjourney for visual design and product ideation.
• A small manufacturer in Italy uses AI tools to optimise production analytics and automate workflows, supported by public policy programmes.
Typical use:
• Marketing
• Operations
• HR and process optimisation
AI Explorers
Profile:
SMEs developing customised AI solutions or AI agents, often in data-intensive or internationally exposed sectors, but still limited to specific functions.
Real-world example:
• A micro B2B trading company in Japan builds custom AI agents to manage multilingual negotiations and customer inquiries, reducing time zones and language barriers for global trade.
Typical use:
• Custom automation
• AI agents
• Domain-specific tools
AI begins to reshape how the firm delivers value.
AI Champions
Profile:
SMEs that embed AI across core operations and strategic decision-making, supported by strong data infrastructure and skilled teams.
Real-world examples:
• A Canadian healthcare technology SME uses AI for clinical documentation, data analysis, and internal workflows, supported by cloud infrastructure and public innovation funding.
• A UK biotech company uses proprietary AI models for drug discovery while also deploying AI tools company-wide for coding, research, and internal knowledge access.
Typical use:
• Enterprise-wide AI
• Strategic decision support
• AI-driven products and services
5. What enables AI adoption in SMEs
The OECD identifies four essential enablers that determine whether SMEs can move along this taxonomy:
1. Connectivity (reliable, high-speed internet)
2. AI-enabling inputs (data, software, compute)
3. Skills (digital and AI-related)
4. Finance (to support experimentation and scaling)
Weaknesses in any of these areas can stall adoption, especially for smaller firms and those in rural regions
OECD Discussion Paper