Decentralized Intelligence as the Leitmotif: What the Industry
10 min read
41 percent of German companies are now using AI-up from 17 percent a year ago. Yet 58 percent of SMEs have no dedicated AI budget, 54 percent don’t know which use cases matter, and more than half of all GenAI projects are shelved after the pilot phase. What’s holding mid-sized firms back, where to start, and which companies are already getting it right.
Key Takeaways
- Adoption doubles: 41 percent of companies with 20+ employees now actively use AI; only 11 percent say it’s not on the agenda (Bitkom, 2026).
- No budget, no plan: 58 percent of SMEs lack a dedicated AI budget. 54 percent don’t know which use cases apply to them (Maximal.digital, 2025).
- Pilot purgatory: Over 50 percent of GenAI projects are discontinued after the test phase-cited reasons: unclear business value, poor data quality, spiralling costs (Gartner, 2025).
- Skills gap: 79 percent of firms lack the necessary AI skills; only 21 percent of SMEs run structured upskilling programs (Stifterverband/McKinsey, 2025).
- Data choke-point: 76 percent struggle with sub-par data quality, 71 percent with data silos across systems (Maximal.digital, 2025).
The adoption gap is closing, but the execution gap remains
The Bitkom study released in spring 2026 reveals striking momentum: 41 percent of companies with 20+ employees are actively using AI, up from 17 percent a year earlier. Another 48 percent are planning or discussing deployment. Germany’s Federal Statistical Office had recorded a 20 percent rate at the end of 2024. Adoption is accelerating faster than most forecasts predicted.
Yet beneath the surface, the picture is less rosy. According to the IW Köln think-tank, only 6 percent of firms apply AI across multiple business units. Just 13 percent invest in paid AI applications, and a mere 3.6 percent develop their own solutions. The AI-Monitoring report from the Bavarian Research Institute for Digital Transformation confirms: roughly one-third use AI, but only about 9 percent have fully implemented it.
Sources: Bitkom, 2026 / Maximal.digital, 2025
The gap between adoption rates and implementation depth explains why so many mid-sized firms still feel they’re at the starting line, even as headlines trumpet otherwise. Testing is not using. Using is not scaling.
Why mid-sized companies hesitate: Five concrete hurdles
1. Unclear business value. According to Maximal.digital, 54 percent of SMEs don’t know which AI use cases are relevant to their business. 81 percent do not systematically measure AI ROI. The problem isn’t the technology; it’s the missing link between AI potential and a concrete business problem. As long as AI is treated as an IT project rather than a business decision, the investment rationale is missing.
2. Missing skills. A joint study by Stifterverband and McKinsey finds that 79 percent of companies lack the necessary AI skills. Only 21 percent of SMEs have a structured upskilling program. Germany currently lacks 109,000 IT specialists, according to Bitkom, and 79 percent of companies expect the shortage to worsen.
3. Data as the bottleneck. 76 percent of SMEs struggle with inadequate data quality. 71 percent report data silos between systems. 83 percent have no comprehensive data strategy. AI needs structured, accessible data. Without a clean data foundation, even the best models fail against the reality of grown IT landscapes.
4. Regulatory uncertainty. 53 percent of companies cite legal hurdles as the main obstacle. 56 percent see more disadvantages than advantages in the EU AI Act. On 2 August 2026, high-risk obligations kick in. The uncertainty is understandable, but waiting is not a strategy: those who introduce AI now can bake compliance in from day one. Those who retrofit later pay twice.
5. Cultural resistance. 67 percent of companies report employee resistance to AI. 58 percent cite fear of job loss as an adoption barrier. Change management isn’t the side dish to an AI project; it’s the prerequisite for it to work.
Why so many AI projects fail
Gartner estimates that over 50 percent of all GenAI projects are shelved after the pilot phase. For agent-based AI, Gartner forecasts a dropout rate exceeding 40 percent by the end of 2027. The causes are rarely technical: runaway costs, unclear business value, poor data quality, and missing risk controls top the list.
The Maximal.digital study offers the SME perspective: 63 percent of SMEs that launched AI projects experienced cost overruns. The average overrun was 34 percent. Only 23 percent of SMEs completed an AI project successfully, even though 86 percent recognize AI as relevant.
86 percent of companies see untapped AI potential in their own operations. Yet only 21 percent of SMEs have a structured program to unlock that potential. Stifterverband / McKinsey, AI Skills Study, January 2025
The pattern is clear: most projects fail not because of the tech, but because of the preparation. Launching without a clear business case, clean data, and clear ownership produces expensive pilots with no future.
What successful mid-sized companies do differently
Some companies prove that AI works in mid-sized businesses. Their common denominator: they don’t start with the technology but with the business problem.
Würth developed the AI assistant “Pico,” a system used daily by 2,800 field sales employees. In January 2024 alone, Pico saved the inside sales team 110 working days-equivalent to roughly five full-time positions. The key: Würth didn’t roll out a generic AI tool but built a solution for a specific operational pain point-relieving inside sales of routine requests from the field.
Maschinenfabrik Reinhausen in Regensburg has used self-learning AI for energy management in its production halls since August 2021. The system now autonomously controls heating, cooling, and ventilation and is also applied to quality assurance and anomaly detection in manufacturing. Reinhausen shows that AI in mid-sized companies doesn’t have to be a future project-it can run productively for years.
The numbers back the approach: according to the 2025 AI Compass study, 71 percent of companies using AI report concrete efficiency gains. Top use cases are in marketing (69 percent), research and development (24 percent), customer service (22 percent), and production (21 percent).
Five steps to get started
1. Start with the business problem. Don’t ask “What can AI do?” but “Where are we losing money, time, or quality?” The most successful AI projects solve concrete operational issues: routine tasks in service, forecasts in procurement, quality control in manufacturing.
2. Audit your data foundation. AI is only as good as the data it receives. Before any AI project: identify data silos, assess data quality, and clarify responsibilities. This doesn’t have to be a major initiative, but it must happen.
3. Start small, measure fast. One use case, one team, one measurable goal. Maximal.digital estimates the typical break-even for AI projects at four to six months. If no measurable impact appears after six months, either the wrong use case was chosen or the implementation wasn’t set up correctly.
4. Build internal skills. Don’t wait for perfect AI experts to appear on the job market. Upskill existing staff. According to PwC, employees with AI skills earn 56 percent more than their peers. Companies that don’t upskill risk losing top talent to employers who do.
5. Plan for compliance. View the EU AI Act not as a brake but as an architectural framework. Companies that introduce AI systems today while embedding documentation, risk assessments, and transparency from day one will avoid costly retrofitting after August 2026.
Bottom line
Mid-sized companies aren’t hesitating because they underestimate AI. They’re hesitating because the bridge between insight and execution is missing: no budget, no data, no skills, no clear ownership. The companies that close these gaps achieve measurable results. Those that wait watch the gap widen. 2026 isn’t the year you must start with AI-it’s the year waiting becomes a competitive disadvantage.
Frequently Asked Questions
How many German SMEs are using AI?
41 percent of companies with 20 or more employees are actively using AI, up 17 percent from the previous year. However, only 6 percent have implemented AI across multiple business areas, and just 9 percent have fully rolled it out.
Why do so many AI projects fail?
According to Gartner, over 50 percent of all GenAI projects are abandoned after the testing phase. The main reasons are non-technical: unclear business value, poor data quality, spiraling costs, and a lack of risk controls. In SMEs, 63 percent of companies experienced cost overruns averaging 34 percent.
What’s the biggest hurdle for AI in SMEs?
Three barriers dominate: 58 percent lack an AI budget, 79 percent lack the necessary skills, and 76 percent struggle with inadequate data quality. The challenge is therefore organizational, not technological.
Which SMEs are using AI successfully?
Würth saves 110 working days per month in its internal services with the AI assistant Pico. Maschinenfabrik Reinhausen has been using AI to manage energy and quality control since 2021. Both companies deployed AI to solve concrete operational problems, not as abstract innovation projects.
How quickly do AI projects pay off?
The typical break-even point is four to six months. Potential cost savings from automation range from 18 to 35 percent, with productivity gains of 22 to 41 percent. The key prerequisite is a clearly defined use case with measurable objectives.
How many IT professionals does Germany lack?
Germany currently faces a shortage of 109,000 IT professionals, according to Bitkom. By 2028, the gap is expected to widen to 768,000. Employees with AI skills already earn 56 percent more than their peers without AI expertise.
What does it cost SMEs to get started with AI?
Getting started doesn’t have to mean launching a major project. AI features already included in existing software such as SAP, Salesforce, or Microsoft 365 are often part of the license fee. For standalone projects, companies typically budget three to six months for a pilot phase and an initial investment starting at 20,000 to 50,000 Euro.
Editor’s Picks
Source of cover image: Pexels / fauxels
Image source: Pexels / fauxels
