What companies actually pay for AI: spending data instead of surveys
Surveys measure intent, transactions measure decisions. What payment data reveals about AI budgets, and why it only goes so far for European mid-market firms.
Founder & Managing Director
Founder and Managing Director of netzstrategen GmbH, on board since 2006. His focus: measurement, analytics and strategy definition. Today above all building AI Operations, from strategy to day-to-day operations. Industry experience in pharma, automotive and manufacturing.
Surveys measure intent, transactions measure decisions. What payment data reveals about AI budgets, and why it only goes so far for European mid-market firms.
Surveys report 88 percent AI adoption, official statistics report 20 percent. Both are correct: they answer different questions. What that means for planning.
Studies put Germany's AI potential at 440 billion euros. At the same time German companies use AI more broadly than others, and see the return later.
Enterprise use is dominated by automation, consumer chat by collaboration. What that gap reveals about where AI actually creates value.
Mid-sized companies use AI less than large ones, and less than small ones. Why the middle is at a disadvantage and what hidden champions can do about it.
77 percent associate AI with progress, 55 percent fear a loss of control. That split sits inside every workforce, and it decides whether an AI rollout works.
62 percent of organizations experiment with AI agents, yet no business function sees more than 10 percent scaling them. The brake is governance, not technology.
BCG surveyed 11,749 workers. Nearly half spend more time managing AI than doing their actual job. Why that is a design flaw, not a training gap.
The Stanford AI Index 2026 shows a shift: inaccurate output is now cited as a bigger risk than security incidents. What that means for running AI in production.
German companies deploy AI more broadly than almost anyone else. The return still takes years to appear. Why that is a structural problem, not a tooling problem.
GPU or cloud? Ollama, vLLM or Open WebUI? An honest cost comparison and decision guide for companies considering running their own LLM.
Which open AI models exist, what they can do, and how they stack up against Claude, ChatGPT & Co.? A factual overview for companies.
EU AI Act, GAIA-X and European AI alternatives: what digital sovereignty really means in the AI era, and which measures actually help.
The AI Act transparency obligations have applied since 2 August 2026. Which four cases are covered, when the editorial exemption applies, and what disclosure looks like in practice, with wording examples.
What happens to company data in AI systems, which legal frame applies (GDPR, Cloud Act, Schrems, EU AI Act), and the three levers for safe use, assessed in one compact overview.
The full process behind this hub: five handoffs between human and machine, the routines driving them and one article as a worked example, documented openly, tools and numbers included.
Microsoft Copilot brings AI straight into Word, Excel, Teams, and Outlook. What it does, what it costs, and what EU data owners need to know about the Flex Routing change.
Gemini lives inside Gmail, Docs, and Sheets, no extra license required. What Google Gemini delivers for business, and what applies to data processing in Workspace.
ChatGPT is the world's most-used AI tool. What GPT-5.5 actually delivers, what ChatGPT Enterprise costs, and where the limits are for business.
Constitutional AI, strong analytical capabilities, and transparent data privacy: what sets Anthropic and Claude apart for business use, and where the limits are.
Claude, ChatGPT, Gemini, Copilot & Co. compared: strengths, data privacy and costs. Which AI tool fits which job in business use.
AI costs exploding? Not with us. Token-Smart by design means every token has a purpose. How to cut AI cost without losing quality, and lift AI ROI.
Managed Machine Mode: when AI systems are no longer triggered by hand but run tasks on their own: monitored, measurable, controlled. How to move from AI assistant to AI coworker.
A practical guide: which settings at the organizational level, what every user must know, and which traps undermine AI data privacy, step by step.
No agent publishes on its own. AI governance is not all-or-nothing. It is autonomy with control as a system, designed function by function.
Building trust in hybrid processes and helping employees find their role in Human+AI systems: the two leadership questions most organizations leave unanswered.
Poor data quality costs companies $12.9M per year on average (Gartner). Why data quality decides your AI ROI, and how to fix it before it scales.
ChatGPT, Claude, Gemini, Copilot: who trains on your inputs, who doesn't? The decisive difference sits between consumer and business tiers. A sober comparison.
For years we delivered outputs, excellent campaigns. But we built no permanence. Why we now act as an AI operations partner, building machines that always deliver.
“Capitalize” has two meanings. The economic one decides: does the AI budget evaporate as an expense, or does a lasting value remain?
The technology is ready, the people are not. The People-Process Gap is the most common cause of failed AI rollouts. Here is how to close it.
AI data privacy risks assessed soberly, by likelihood and by case. What really counts, what is overstated, and what companies must keep on their radar.
The difference between AI systems people use and those left on the shelf: whether they were built with the team or for the team.
Two talks from Business Forum 2026, one diagnosis: AI is not a tool question. It changes how companies are organized inside, and whether they are found at all from the outside.
88% of companies use AI, only about 5% create value at scale (BCG, 2025). The difference isn't the model. It's scaling: consistency, cockpits, standards.
Why do well-meant AI rollouts fail? Because they address one of five areas and ignore four. The AI Operations Framework shows what must work together.
'It saves 40% of the time' is not a business case. A real AI operations ROI has three pillars, a payback period, and an uncomfortable answer to: what does doing nothing cost?
Most agencies hide their process behind a contact form. We don't. Here is what AI operations onboarding looks like: step by step, with clear costs and an exit at every point.
A $965B valuation for Anthropic, and 90% of companies without a measurable productivity effect (NBER, 2026). The Implementation Gap sits in between.
61% of companies have not moved beyond pilot projects (McKinsey, 2025). Why the Pilot Graveyard is not a technology problem, and how to avoid it.
Production from Day One is an architecture decision, not a promise. Starting AI as an experiment yields an experiment. Here is how AI ships into real operations.
AI Operations describes the efficient and durable operation of structures and processes that run through the use of AI in a hybrid organization of humans and machines.
The most common pattern in failed AI projects: buy the tool first, rethink the process later. Why Workflow-First reverses the order, and lifts ROI.