Generative AI statistics 2026

GenAI spending by segment (US$ millions)

Hardware dominates: devices and servers together account for about 90% of 2025 GenAI spending (our calculation from the Gartner table). Software ($37.2 billion) and services ($27.8 billion) are much smaller, but grow fast.

Note that these are different measures: Gartner forecasts spending (including AI-capable PCs and smartphones), Bloomberg Intelligence (June 2023) forecasts revenue of the generative AI market. Do not compare the two directly.

Sources: Gartner, 2025, Bloomberg Intelligence, 2023

AI use by company size (Netherlands)

Marketing is the top use case: 35% of Dutch companies that use AI apply it to marketing or sales (CBS), and 86.4% of marketing teams worldwide use AI in at least a few areas (HubSpot, 2026), often with an AI marketing agency leading the rollout.

Among developers, 84% use or plan to use AI tools (Stack Overflow, 2025). Worldwide, 70% of organizations use generative AI in at least one business function (Stanford AI Index, 2026). Small companies lag behind large ones; SMBs tend to catch up fastest with practical AI marketing for SMB playbooks.

Sources: CBS, 2025, HubSpot State of Marketing, 2026, Stack Overflow Developer Survey, 2025, Stanford AI Index, 2026

What the studies measured

Important nuance: exposure is not the same as job loss. The ILO finds that gen AI exposure is far higher in high-income countries (34% of employment) than in low-income countries (11%), and mostly affects specific tasks within occupations.

The biggest measured gains go to less experienced workers: in the customer support study, novices gained 34% versus 14% on average.

Sources: ILO, 2025, NBER, 2023, GitHub, 2022, Stack Overflow, 2025, Eurostat, 2025

Open source is losing enterprise share: open models handled 11% of US enterprise LLM usage in 2025, down from 19% a year earlier (Menlo Ventures). Hugging Face nevertheless lists 3,225,742 models.

The performance gap widened slightly: in March 2026 the best closed model led the best open model by 3.3%, up from 0.5% in August 2024, and six of the top ten Arena models are closed (Stanford AI Index 2026). In the Netherlands, 18.2% of companies (10+) use ready-to-use commercial AI and 8.1% adapt open-source AI themselves (Eurostat 2025).

Sources: Menlo Ventures, 2025, Hugging Face, 2026, Stanford AI Index, 2026, Eurostat, 2025

EU AI Act: key obligations for generative AI

• Transparency: AI-generated content must be labeled as such. Providers of generative AI systems placed on the market before 2 August 2026 must comply with the marking rule (Article 50(2)) by 2 December 2026.
• Copyright compliance: Training data must be documented. Providers must publish a summary of the training data used.
• Risk classification: Generative AI models fall under 'general-purpose AI' (GPAI), with extra obligations for models with systemic risk.
• Deepfakes: Synthetic audio, video and images must be disclosed as artificially generated.
• Cybersecurity: GPAI models with 'systemic risk' must undergo adversarial testing (red-teaming).

EU AI Act timeline

Legal uncertainty is a brake: of Dutch companies that considered AI but did not adopt it, 49% cite privacy and 42% legal consequences (CBS, 2025). Measured over all companies, unclear legal consequences hold back 3.8% of Dutch firms versus 5.9% across the EU (Eurostat, 2025).

Sources: AI Act, Article 99, AI Act implementation timeline, CBS, 2025, Eurostat, 2025

The conclusion is clear: generative AI is no longer a trend but a fundamental shift in how businesses operate, with real differences in quality, cost and regulation to manage. Most modern teams build a proper AI marketing stack to integrate generative AI across content, advertising, and automation workflows.