6 Trends, die Innovation in Unternehmen
2027 neu prägen
KI hat den Output von Innovationsteams vervielfacht – aber nicht die Zahl der Menschen, die entscheiden, was weiterverfolgt wird. Unser dritter Jahresreport zeigt, wo die Entscheidungskapazität an ihre Grenzen stößt und was Teams anders machen, die diese Lücke schließen.
Discuss this report with your AI:
Choose your AI to explore this report, or copy the prompt.
Foresight moves from inspiration to insurance
Foresight has always had to fight for business relevance. It used to be sold as inspiration: a picture of how the world might develop. Now it’s funded as protection against being outpaced, and the organizations that arrived with budget this year had a specific exposure behind them. One had been repeatedly late to a market shift. Another was caught cold by developments from China.
Most foresight practices sit mid-transition, and the open question is where foresight belongs once someone agrees to pay for it. Two models are forming. On its own, either one pays only half.
of 742 organizations, the largest trend cluster this year
- Foresight moves from inspiration to insurance46% (342)
- Innovation output outruns its decision rules40% (294)
- Funding questions outgrow the spreadsheet23% (173)
- Cross-unit visibility demands a mandate23% (170)
- Institutional memory decides survival22% (163)
- Security takes the AI decision15% (112)
Source: ITONICS · 2,187 conversations, September 2025–August 2026. Organizations may appear in several areas; shares do not sum to 100%. Percentages are rounded.
“Scanning used to take most of the effort, and agentic AI has genuinely changed that.”
Innovation output outruns its decision rules
AI compressed drafting, scanning and synthesis. Campaigns scale at almost no cost, and business units launch their own initiatives instead of queuing for the center. But the stage that decides what proceeds didn’t get faster.
Drug discovery shows the pattern: AI-discovered molecules far outperform the historic rate in Phase I, then perform no better in Phase II, where programs die. At low volume, a person’s judgment held the process together. At today’s volume, the missing rules become visible. Automation can carry the new load, but it can’t supply rules nobody has written down.
Fewer than 1 in 5 first conversations (<20%) showed an owned, gated process
“Process work usually stalls when basic questions remain unanswered: how should the process work, how should it be governed, and who does what, decides what, and when?”
Funding questions outgrow the spreadsheet
Innovation budgets have always needed defending. What changed is the standard of the defense. A good case for the next project used to be enough. Now the request has to show what the last one delivered, or why it was stopped.
Meanwhile, the bigger picture isn’t moving: global corporate R&D investment grew 6.3% in 2024, while international patent filings have barely shifted since 2020. Most organizations put off fixing the portfolio until a contract renewal or budget round forces it. By then, what gets cut is whatever is easiest to stop.
Portfolio reviews
1 in 4
organizations rebuild their portfolio by hand for every review.
“More ideas do not by themselves produce better outcomes. What was missing was an owner, a mandate, a decision, or the evidence to defend one.”
Security takes the AI decision
Innovation teams used to choose their own tools, run a pilot, and ask for budget once it worked. That sequence is gone. Information security, IT risk and legal now arrive with the decisive questions at scoping, sometimes leading the search themselves. At one electronics manufacturer, nobody from innovation or R&D attended the demo.
Meanwhile, about a third of organizations declined to buy software because coding agents made building it viable. Teams split almost evenly between AI already running and AI still being planned, and the two halves need very different next moves.
Share of themes in the security-led AI cluster; not percentages of all 742 organizations.
“The innovation team still leads, and it now decides together with security and IT. I see that as an advantage.”
Cross-unit visibility demands a mandate
In 2025, centralizing the portfolio was the top priority. In 2026, the answer was unified innovation PMOs. Three years on, business units are still asking to be made visible to each other. Innovation swung decentralized, then centralized, and has landed in the middle.
Software for a group-wide view has existed for years, and the gap has outlasted it. What organizations describe as one “central vs. local” problem is actually three different ones, each with its own remedy. Applying the wrong one costs a year.
Innovation blockers
2×
Internal politics are named twice as often as other innovation blockers.
Institutional memory decides survival
In many organizations, the innovation function is one resignation away from starting over. Its mandate and the history of what it tried, stopped and decided often sit with a single person, and they’re rarely written down.
The layer that used to replace that person is thinning too, as AI absorbs the junior work through which people once built judgment. The function is losing memory at both ends, and neither loss shows up in a budget cycle. What can be secured before a leadership transition is often impossible to recover after it.
Innovation continuity
More than
1 in 5
organizations say their innovation function rests on one person.
Continuity at risk.
“The individual builds the process, including the agents that execute the steps along it, and from there the work scales to the team and through the organization.”