1
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
WHITEPAPER WITH 70+ EXPERT STATEMENTS
AI Trends
Report 2026
2
AI Trends Report 2026statworx & AI Hub Frankfurt
2026 – 02 2
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
AI in 2026 – Between breakthrough and stress test
Foreword by Sebastian Heinz, Founder & CEO statworx | AI Hub Frankfurt
2026 will be a decisive year for AI. Artificial intelligence is no longer new, no
longer in need of explanation, and no longer an exclusive topic for technology
departments, startups, or research centers. AI has arrived. In companies, in
products, in everyday processes, and in political debates. The phase of wide-
eyed observation is over. What comes next is a phase of proof.
The much-discussed AI bubble continues to inflate. Whether it bursts with a
bang in 2026 or deflates gradually will become clear in the months ahead.
The global race for AI dominance
The development of AI has become significantly geopolitical over the past
year. The US, China, and Europe are pursuing different strategies that go far
beyond technology and touch on issues of economic power, security, and
global influence.
The United States is intensifying its AI strategy in the global technology race.
Washington is betting on the international diffusion of American AI technology
to secure influence abroad. In 2025, the US government announced plans to
actively promote exports across the entire AI stack and to reassess existing
Foreword
export restrictions. This signals a strategic shift toward pursuing dominance
through technological leadership rather than isolation.
This approach is being reinforced by new AI partnerships with strategically
important countries, including in the Middle East, aimed at embedding stan-
dards early and limiting China’s influence. Domestically, pressure is mounting
to define clear guardrails for AI safety, even as the overall stance remains
innovation-driven. At the same time, AI is gaining importance as a national se-
curity issue.
China, in turn, is stepping up its efforts in 2026 to become more technolo-
gically independent and underpin its global leadership aspirations. Against
the backdrop of Western export controls, Beijing is investing heavily in its own
chip development and production capacities. At the same time, Chinese AI
companies are focusing on open-source models in order to help shape the
global AI infrastructure. This strategy is proving effective: Chinese models are
increasingly being used outside the country.
Geopolitically, the technology conflict between East and West remains tense,
with both sides likely to continue pursuing so-called decoupling in 2026. In AI,
HELLO WORLD
3
AI Trends Report 2026statworx & AI Hub Frankfurt
2026 – 02 3
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
HELLO WORLD
the fault lines extend beyond chips and software to include raw materials.
China has secured access to rare earths in South America, giving it control
over critical resources for AI hardware - a move that could prompt counter-
measures from the United States. At the same time, Beijing is increasingly le-
veraging AI as a tool of geopolitical influence, from disinformation to cogniti-
ve warfare, while simultaneously asserting digital sovereignty.
Europe is charting its own course in AI. With the AI Act, the EU established an
early regulatory framework that positions Europe as a global standard-set-
ter for trustworthy AI. At the same time, awareness is growing that Europe is
falling behind economically. It lacks comparable players both in large-scale
AI platforms and in the semiconductor sector. As a result, initiatives aimed at
technological sovereignty are moving to the forefront, including the build-out
of high-performance computing infrastructure and a European AI stack.
AI is also gaining importance in security and defense policy, reflected in rising
investment levels. Internationally, Europe is promoting its values-based, risk-
oriented approach, but faces fundamentally different governance models in
the US and China.
For Europe, 2026 will therefore be a balancing act. It seeks to set standards
without falling further behind economically. Low AI adoption across enterpri-
ses, talent shortages, and the difficulty of scaling excellent research remain
key challenges. Germany is a prime example of this tension between scienti-
fic strength and industrial execution.
Investments and capital
AI also dominated the venture capital landscape in 2025: more than half of all
global VC funding flowed into AI deals. Expectations for 2026 are more mixed.
On the one hand, there is still a vast amount of dry powder waiting to be de-
ployed - large funds have explicitly focused on AI and are searching for the
next OpenAI. As a result, select companies continue to command extreme
valuations, as OpenAI and Anthropic demonstrate.
On the other hand, skepticism is growing. Many investors view current va-
luation levels as difficult to sustain over time. With interest rates remaining
elevated and geopolitical risks persisting, financing is likely to become more
challenging, especially for very early-stage startups. As early as 2025, capital
was already heavily concentrated among a small number of large players, whi-
le early-stage ventures lost visibility.
In 2026, this “flight to quality” could intensify: venture capital is likely to flow
preferentially into AI infrastructure and into application-focused startups
with a clear industry focus. Generic AI offerings without a distinct competitive
edge are coming under increasing pressure.
Still, the AI sector remains one of the few bright spots in the broader tech
landscape, meaning venture capital remains abundant by historical stan-
dards. Regional disparities may deepen, however. In 2025, North America
4
AI Trends Report 2026statworx & AI Hub Frankfurt
2026 – 02 4
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
HELLO WORLD
accounted for roughly 70 percent of global AI VC investment. Europe trails sig-
nificantly behind, which could prompt more European startups to relocate to
the US in 2026 in search of funding.
Quo vadis, Germany and Europe?
Global AI developments in 2026 pose particular challenges for Europe - and
Germany in particular - while also opening up new opportunities. On the one
hand, a massive future market is emerging that Europe must tap into. On the
other, Europe risks being confined to the role of consumer or regulator if no
comparable homegrown AI players emerge. Germany, as the EU’s leading eco-
nomy, feels this pressure acutely: while US companies such as OpenAI and
Nvidia and Chinese players like Baidu and Huawei are delivering global AI plat-
forms and hardware, Germany (and the EU more broadly) lacks an equivalent
counterpart.
This technological gap deepens economic dependencies. German compa-
nies are already heavily reliant on US cloud providers to develop and operate
AI systems. As a result, interest in European digital and AI sovereignty is high,
even though its impact will only materialize over the medium term. In the short
run, Germany will continue to import most AI technology while attempting to
remain globally competitive in select niches, such as industrial automation.
Globally, enormous capital continues to flow across the AI value chain. The US
and China view AI as a core driver of growth and power, while Europe has so far
been unable to mobilize comparable levels of investment. In some cases, in-
dividual US corporate investments exceed the EU’s total public spending on AI.
Germany therefore needs to find new paths to remain competitive. These in-
clude deeper industrial collaboration, tighter integration between research
and industry, and stronger talent retention. If successful, AI can make Germa-
ny’s industrial base more efficient and flexible. If not, the country risks a fur-
ther erosion of its competitive position. Accordingly, German companies face
the task in 2026 of establishing AI as a core capability and aligning business
models, processes, and products with an AI-driven future.
The year 2026 will determine what role Europe plays in the global AI order. With
its human-centered, risk-based approach, the EU has already made an im-
pact. The so-called “Brussels Effect” is evident as countries such as Brazil
and Canada adopt elements of the AI Act, and even China selectively aligns
with certain European regulatory concepts. This strengthens Europe’s inter-
national soft power.
At the same time, without democratic oversight, such regulatory frameworks
can also be misused to restrict freedom. Europe must therefore closely mo-
nitor how its standards are applied internationally. For Germany, this means
further strengthening its mediating role in international forums, for example
in standard-setting and multilateral partnerships.
Global AI trends could prompt Europe to invest more strategically in 2026. The
year will become a litmus test for Europe’s ambition to be a rule-maker rather
than a rule-taker. If it succeeds in balancing innovation and regulation, Euro-
pe can pursue an independent path between the US and China. If not, it risks
fading into irrelevance as a mere user of foreign AI platforms. The decisions
made in 2026 will thus be critical to Europe’s economic and political future.
5
AI Trends Report 2026statworx & AI Hub Frankfurt
2026 – 02 5
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Outlook: AI between promise and responsibility
The AI Trends Report 2026 portrays a technology at a turning point. The phase
of uncritical optimism is over. Instead, more fundamental questions are co-
ming to the forefront: How do we ensure that AI truly creates value? How do
we strike the right balance between innovation and regulation? And how do
we prevent the AI revolution from exacerbating existing inequalities between
regions and social groups? These questions underscore the breadth of AI’s
impact on our economy, society, and environment.
The following chapters offer guidance for decision-makers in business, poli-
tics, and society. Because one thing is clear: the decisions made in 2026 will
shape AI development, and thus the future of our societies, for years to come.
I hope you enjoy reading the AI Trends Report 2026 and find it thought-
provoking!
HELLO WORLD
Sebastian Heinz
Founder & CEO statworx | AI Hub Frankfurt
6
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Content
2025 Review
How accurate were last year’s predictions on the most important
AI trends – and what has happened since?
7
19 Category 1: Company
Governance, oversight, and compliance as the foundation for
scalable AI systems
Category 2: Work, Organization & Communication
AI-driven transformation: connected workers, AI agents,
and new interfaces
50
Category 3: Economy, Regulation & Power
AI in global competition: infrastructure, energy, and regulation
as key drivers of success
75
Category 5: Technology
The next technological leap in AI: simulation, structure,
and integration
112
Category 4: Science, Culture & Society
AI as a catalyst for scientific research, synthetic media,
and co-creation
92
Outro
How to translate the analyses and insights from this
AI Trends Report into action
133
Introduction
An overview of the objectives and five thematic focus areas of
the AI Trends Report 2026
13
About statworx & AI Hub Frankfurt Rhein Main
Meet the team: Learn who is behind this report
and what drives our work
136
Call for Experts 2027
Help shape the conversation and become a contributor
to the AI Trends Report 2027
135
REVIEW
Every year, we open our AI Trends Report by looking back at the forecasts
from the previous year. For the AI Trends Report 2025, we formulated 16 core
theses on the future of AI.
One year later, the picture is clear: some of these predictions materialized
faster than expected, others remain more aspirational for now – and a few
were overtaken by reality at record speed. This retrospective not only high-
lights how rapidly developments in AI are accelerating, but also which pat-
terns are becoming firmly established and which trends will be especially
relevant over the next twelve months.
2025
8
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
AI agents made meaningful progress
in 2025, driven in part by Gemini
and . At the same time, examples
such as Klarna’s rollback of customer
service automation highlight the still
limited maturity of current agent so-
lutions. As a result, a breakthrough of
fully autonomous AI agents is not ex-
pected before 2026+.
TREND 1
AI Agents revolutionize the
job market
Tools such as Copilot, Cursor, and mo-
dern AI editors significantly increased
productivity, but broad-based demo-
cratization failed to materialize. The
main barriers remain a lack of AI skills,
governance, and process maturity. For
developers, these tools are strong
productivity boosters – for citizen de-
velopers, their impact remains limited.
Tooling is evolving faster than organi-
zations can adapt.
TREND 2
Low-code and no-code
democratize software
development
AlphaFold 3, GNoME, and GenCast sig-
nificantly accelerated scientific re-
search and, in some cases, outper-
formed human experts. However, a
true autonomous breakthrough failed
to materialize - such as a top-tier pu-
blication initiated and authored ent-
irely by AI. As a result, 2025 was a year
of rapid progress, not of major, inde-
pendent discovery.
TREND 3
AI achieves its first big
scientific breakthrough
TREND REVIEW 2025 | INNOVATION & TRANSFORMATION
Medium – progress, but no breakthrough
Hit rate
High, with clear limits
Hit rate
Partially confirmed
Hit rate
9
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Apple, Meta, OpenAI, and Tencent sig-
nificantly curtailed AI functionality in
Europe. While the AI Act set ambitious
standards, it also introduced delays
and limited feature rollouts. Europe is
regulating effectively, but at the cost
of technological momentum. The gap
with the US and China is widening.
TREND 4
Tech giants release “AI light
versions” for the EU market
Despite the R1 shock, stock pullbacks,
and broader market corrections, no
crash materialized. Capital continued
to flow, albeit more selectively and
with a stronger focus on sustainable
business models. 2025 delivered con-
solidation rather than collapse. The
real stress test may not come until
2026.
TREND 5
The AI investment bubble
bursts
Tools such as HeyGen, Pika 2, and Veo
3 established avatars across marke-
ting, training, and communications.
At the same time, deepfakes, identity
misuse, and copyright issues increa-
sed sharply. Progress and risk are ri-
sing in parallel. Rules and safeguards
remain largely absent, even as adop-
tion accelerates rapidly.
TREND 6
AI avatars shape new crea-
tive and ethical standards
TREND REVIEW 2025 | REGULATION & INVESTMENT
Fully validated
Hit rate
Hit rate: Did not play out
Hit rate
Hit rate: Very high
Hit rate
10
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
The obligation to ensure AI compe-
tence was introduced, but implemen-
tation remained limited. Studies point
to significant literacy gaps and low
employee engagement. Article 4 pro-
vided momentum, but failed to trigger
structural transformation. AI compe-
tence remained a central weakness
in 2025.
TREND 7
Article 4 of the AI Act
promotes AI education in
companies
Duolingo Max, NotebookLM, and Tea-
chAI demonstrate clear potential, but
a broad-based breakthrough failed
to materialize. Educators are over-
burdened, strategies are lacking, and
meta-skills are rarely taught. Pro-
gress is occurring in isolated pockets
rather than systemically. Education
remains a slow-moving domain of
transformation.
TREND 8
Automated AI learning
platforms democratize
education
GPT-4o and Gemini Ultra established
dialog-based AI interaction. Promp-
ting increasingly evolved into context
engineering - the design of intents,
context, and workflows. Users began
interacting with AI systems in a more
natural and continuous way. In 2025,
the shift from commands to conver-
sations became a reality.
TREND 9
Conversational AI replaces
prompting
TREND REVIEW 2025 | EDUCATION & DEVELOPMENT
Formally confirmed – weak in practice
Hit rate
Partially achieved
Hit rate
Very high
Hit rate
本报告来源于三个皮匠报告站（）,由用户Id:619989下载,文档Id:1146834,下载日期:2026-03-06
11
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
AI became a standard feature across
operating systems and applications.
Windows Copilot, Gemini in Android,
and Apple Intelligence are shaping
interaction and automation. Nvidia Di-
gits and AI PCs further reinforced this
trend. The forecast proved accurate
– but turned out to be more conser-
vative than necessary.
TREND 10
AI integration transforms
user experiences
Models such as OpenAI’s o3 series
and reasoning models like DeepSeek
R1 represented new performance di-
mensions in 2025, complemented by
advanced families such as GPT-5 and
Gemini-3. Measurable performance
gains are clearly evident, yet the gap
to true AGI remains substantial. The
second wave of LLM development is a
reality.
TREND 11
Instead of a plateau, we
see further progress in LLM
performance
Approaches such as OpenAI Opera-
tor, Claude Computer Use, and devi-
ces like the rabbit r1 illustrate how AI
is increasingly operating software in
a human-like way. At the same time,
terminology, reliability, and safety
standards remain inconsistent. The
vision is clearly visible – but maturity
levels vary widely. Trust remains the
central hurdle.
TREND 12
LAMs and CUAs take
control of your desktop
TREND REVIEW 2025 | TECHNOLOGY & PROGRESS
Fully validated
Hit rate
Fully validated
Hit rate
High, but fragmented
Hit rate
12
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Large-scale government
initiatives are making litt-
le progress. While private
players such as Microsoft,
Deutsche Telekom, and the
Schwarz Group are investing
in data center and cloud in-
frastructure, a globally com-
petitive AI hub has yet to
emerge. Initiatives exist, but
remain fragmented. Germa-
ny’s structural dependence
on Big Tech persists. 2026 will
be a decisive year.
TREND 13
Germany plans an AI
data center
Black Forest Labs, DeepL,
and Helsing made progress
and attracted capital, but a
global champion has yet to
emerge. Infrastructure gaps,
talent shortages, and limi-
ted VC firepower continue to
hold Europe back. Europe is
becoming more visible – but
not leading.
TREND 15
A German AI startup
achieves a global
breakthrough
Companies with clear gover-
nance structures are bene-
fiting: lower risk, stronger
compliance, and faster roll-
out of AI solutions. Globally,
regulation remains fragmen-
ted, and bias and alignment
issues remain unresolved. In
2025, governance became a
strategic imperative.
TREND 14
AI governance beco-
mes a competitive
advantage
TREND REVIEW 2025 | CORPORATES & START-UPS
Efficiency-focused models
such as R1 continue to drive
down usage costs, while de-
velopment, training, and pre-
mium services still require
substantial investment. Ri-
sing compute demands and
more differentiated pricing
tiers are creating a two-tier
market: affordable usage,
yes – affordable develop-
ment, no.
TREND 16
The era of cheap AI
is over
Confirmed, but differentiated
Hit rate
Still open
Hit rate
Clearly confirmed
Hit rate
Not yet achieved
Hit rate
13
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
The AI Trends Report 2026 presents 20 dynamic AI trends, organized across
five core categories:
Intro
WHAT’S IT ABOUT?
Company
Work, Organization & Communication
Economy, Regulation & Power
Science, Culture & Society
Technology
1.
2.
3.
4.
5.
This comprehensive trend analysis combines market observation with in-
sights from leading organizations, offering a well-grounded and practical
assessment of current AI developments. The report helps readers identify
strategic opportunities and position themselves for long-term success wit-
hin a rapidly evolving AI ecosystem.
COMPANY
CATEGORY 1
AI projects are facing
unprecedented
pressure to perform
TREND 1
Results over experiments
Where data and
processes are solid,
agentic AI can reach
its full potential
TREND 3
Agents on probation
Humanoid machines
are entering the
workplace
TREND 4
Robots, ready for work
Clear rules are
shifting from manda-
tory compliance to a
driver of returns
TREND 2
AI governance as a success
factor
AI hardware turns
assistants into cons-
tant companions
TREND 5
The second self
With AI, fewer people
can create far more
enterprise value
TREND 7
One-person unicorn
AI is reshaping
knowledge jobs and
redefining core skills
TREND 8
The new age of knowledge work
AI becomes the
command center
of marketing
TREND 6
The quiet Intermediary
WORK,
ORGANIZATION &
COMMUNICATION
CATEGORY 2
ECONOMY,
REGULATION &
POWER
CATEGORY 3
Policy makers are
pulling back on AI
oversight
TREND 11
Regulation in reverse
Artificial Intelligence
is reshaping the logic
of military power
TREND 12
AI on the battlefield
Whoever controls
power controls the
future of AI
TREND 10
Energy as the bottleneck
Processing power
becomes a question
of global power
Geopolitics of compute
TREND 9
SCIENCE,
CULTURE &
SOCIETY
CATEGORY 4
Universities become
drivers of the new AI
learning culture
TREND 15
Next-level education
Artificial creativity
breaks into the
charts
TREND 16
The first AI hit
AI blurs the line
between reality
and fiction
TREND 14
Real was yesterday
AI becomes the
world’s new R&D
department
Research
TREND 13
TECHNOLOGY
CATEGORY 5
DataOps & AgentOps
bring control and
efficiency
TREND 19
Order in the AI chaos
OpenAI builds the
operating system of
the AI era
TREND 20
ChatGPT goes platform
The biggest AI leaps
are now happening
behind the scenes
TREND 18
Progress in slow motion?
World models
connect simulation
with reality
Learning worlds for machines
TREND 17
19
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02statworx
The frenzy is over. After two years of AI hype and billions in investment,
many companies are now reassessing: production use remains limited,
and measurable results are still scarce. 2026 will therefore be the mo-
ment of truth. AI initiatives must deliver tangible contributions to revenue,
productivity, or cost reduction. Those that fail to do so risk losing both
budget and strategic priority.
Generative AI has sparked unprecedented euphoria. Impressive demos and
shiny prototypes took boardrooms by storm. Budgets for “anything AI” were
readily approved. Hundreds of billions of US dollars have already been pou-
red into technology and infrastructure worldwide, driven by ambitious ex-
pansion plans. But reality often fails to keep pace with the vision. Numerous
studies document high dropout rates and endless pilot projects. Only a tiny
circle of pioneers actually achieve a significant impact. This gap will close in
2026: The phase of experimentation is over.
CLICK HERE!
Even if we lose a couple
hundred billion, it would
suck, but it‘s better than
being behind the race for
super intelligence.
Mark Zuckerberg, CEO Meta Platforms
“
© 2026 statworx & AI Hub Frankfurt
Results over experiments
01 | COMPANY
TREND 1
AI projects are
facing unprece-
dented pressure
to perform
AI Trends Report 2026
20
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
CATEGORY 1: COMPANY – TREND 1
From pilot projects to real return on investment
The data paints a clear picture. Gartner predicts that by the end of 2025,
at least 30 percent of all GenAI projects will be discontinued after proof of
concept, having failed due to unclear business logic, poor data quality, or
exploding costs. An MIT study is even more drastic in its assessment: around
95 percent of the enterprise AI investments examined have not yet delivered
any demonstrable contribution to business results. Only a handful of pilots
translate into balance-sheet relevance.
A survey of 3,700 executives reveals an apparent paradox: while 54 percent
of companies are already reporting positive returns, 62 percent of all pro-
jects remain stuck in the perpetual pilot phase. This pattern is not new. Back
in 2020, long before the GenAI wave, only about one-fifth of companies were
using productive machine learning models, despite rising budgets. The hurd-
le was never in modeling, but always in deployment and scaling. Observers
have coined the apt term “pilot theater” for this phenomenon: a parade of
impressive prototypes that never make it into critical business processes.
At the same time, pressure from finance departments is growing. They now
demand measurable effects from GenAI within one to three years, applying
the same standards that apply to other major transformation programs. The
era of open, technology-driven experimentation is coming to an end. It is
being replaced by AI programs that must be measured against hard business
metrics.
Power shift: Those who deliver stay
The call for profitability is fundamentally changing the power structure wit-
hin companies. CFOs and board members are caught between two conflic-
ting forces: on the one hand, there is growing conviction about the trans-
formative potential of AI, while on the other, evidence of bad investments
is mounting. Maturity studies reveal a clear two-tier society. Of the 2,000
managers surveyed, only about eight percent achieve “front-runner” status,
meaning they belong to the elite group that has successfully scaled seve-
ral strategic AI bets. Their lead is paying off: they are achieving significantly
higher revenue growth and a better return on investment than the eternal
experimenters.
This dynamic shifts the internal balance of power. AI labs and experimental
units without clear business ownership face growing pressure to prove their
value. Where results are lacking or cannot be quantified, the business case
is at risk. In return, departments with demonstrable benefits gain influence.
Teams that shorten processing times, reduce error rates, or generate new
revenue can confidently expand their programs. This forces data, IT, and
business teams to work more closely together. Without clear responsibi-
lities, defined KPIs, and coordinated risk guidelines, projects are at risk of
being scrapped after the proof of concept.
But structures alone are not enough. The bottleneck often lies with the hu-
man factor. A global EY survey shows that while many employees use AI tools,
21
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
only a minority integrate them transformatively into their work. As a result,
companies are wasting an estimated 40 percent of potential productivi-
ty gains, mostly due to a lack of training and unclear guidelines. Economic
pressure is forcing organizations to change their mindset: AI is no longer just
a technology project, but requires a fundamental adjustment of processes,
roles, and skills.
Real gains in core operations
Success stories do exist but they are rarely found in glamorous showcase
projects. Instead, they tend to emerge in the “engine room” of organizati-
ons: core and back-office processes. Invoice and document workflows, for
example, are undergoing a quiet times are shrinking
to a fraction of what they used to be, and costs are falling dramatically. One
provider reports over 90 percent cost reduction in invoice processing and
drastically shortened review cycles in due diligence. Service and workflow
platforms also generate considerable returns through intelligent collabo-
rations, as in the case of ServiceNow and the Japanese technology group
Fujitsu. The reported ROI figures reach six to seven-digit amounts, as freed-
up productivity and additional business value add up over the years.
The same applies to business process outsourcing and risk management,
which benefit from well-integrated AI solutions. According to an MIT ana-
lysis, annual savings in the millions can be achieved when these systems
are integrated into core business processes. The greatest benefits arise
in seemingly “unspectacular” areas such as reconciliation, claims proces-
sing, or standardized documentation. Marketing or sales pilots dominate the
headlines, but often deliver inconsistent results.
To put this power on the road, a new infrastructure is emerging. Governan-
ce-strong “AI Factory” approaches, such as that from Dataiku, merge infras-
tructure, model access, and machine learning operations into end-to-end
workflows. At the same time, no-code frameworks are democratizing the
development of domain-specific AI agents. They make it possible to train
agents on company data, test them automatically, and roll them out in a
controlled manner. This is creating a niche in which several industry giants
are actively positioning themselves. Where AI is consistently docked to sca-
lable processes and built on a solid platform, significant contributions to
results are already a reality today.
WHAT IS AN AI FACTORY?
An AI Factory is a strategic, scalable operating model that enables
organizations to systematically develop, deploy, and continuously
improve artificial intelligence across the entire lifecycle - compa-
rable to an industrial production line, but designed for AI-driven so-
lutions.
Core characteristics of an AI Factory:
• Standardized end-to-end processes for AI
• Modular, scalable data and model architecture
• Clear governance and accountability
• Close collaboration between business units, IT, and Data/AI teams
• Kontinuierliche Optimierung im Betrieb
CATEGORY 1: COMPANY – TREND 1
22
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Separating the wheat from the chaff
This development is framed by the debate about a possible “AI bubble.”
International organizations and economists draw parallels to the dot-com
era: astronomical investments and exploding valuations meet the convicti-
on that the technology will unleash enormous productivity gains in the long
term. Optimistic forecasts assume that AI will significantly increase overall
economic productivity in the coming decade. Provided that integration into
business processes is successful. At the same time, experts warn of over-
heating in some segments, which could trigger a painful market correction.
This leads to three clear imperatives for companies. Projects without a clear
link to the value chain will struggle to secure funding, as the era of experi-
mentation is over. Platforms, standards, and partner networks are becoming
critical success factors for scaling, security, and governance. And finally,
the focus is shifting to people: investing in skills is not a “soft side issue,”
but a hard prerequisite for realizing technical potential in the first place. By
2026, the market will separate the wheat from the chaff: those who ope-
rationalize GenAI will move ahead, while others will have to write off costly
experimentation.
CATEGORY 1: COMPANY – TREND 1
Expert Statements TREND 1
HP leverages AI across its internal operations
to boost productivity, resilience, and employee
experience across the company to define
the future of work. All are part of HP’s broader
transformation into a data-driven, intelligent
enterprise. Furthermore HP deploys AI-Note-
books and uses AI-driven telemetry based on
HP WXP (Workforce Experience Platform) to
enhance employee experience.
Adrian Müller
Vice President and Managing Director
HP Germany/Austria/Switzerland
„
We need to accelerate the change. Start-
ups and scale-ups can be a booster for
incumbents. But this also requires a new mind-
set in our industry. More courage, more agility,
and, not least, strong commitment from top
management.
Peter Stockhorst
Board Member Digital Business &
Partnerships Zurich Germany | CEO DA Direkt
„
Expert Statements TREND 1
Many companies confuse speed with progress:
they experiment a lot but scale little. In the
results-driven age of AI, it‘s not the fastest
who win, but those who systematically inte-
grate AI into decision-making processes
and accountability. The decisive lever is AI
literacy – as an organizational capability to
critically evaluate AI outcomes and use them
productively.
Prof. Dr. Alexander Benlian
University Professor
Technische Universität Darmstadt
„
Deka has built the foundation to scale AI
adoption with business value in 2026:
1. Scalable infrastructure (Company GPTs & RAG)
2. Provision of data (clients, assets, markets)
3. Enablement and skills.
We will improve the client experience, enhance
the quality of our products and increase
efficiency. The expected value of AI has not
changed; however, it is less of a quick win than
expected.
Daniel Kapffer
CFO/COO
DekaBank
Expert Statements TREND 1
2026 is on value delivery: pick E2E processes,
automate them with AI and measure the
impact. Build fast in short cycles, ship to real
users, learn from feedback and prove value
early. Scale what works, stop what doesn’t.
For that you need to democratize AI and have
interdisciplinary teams with business people,
data scientists and data experts. That’s the
game.
The AI bubble won’t burst; it will consolidate.
Many incumbents paid for slow, expensive
internal AI R&D with limited returns, while
startups ran tighter feedback loops and solved
concrete problems end-to-end under real
market pressure. In 2026, companies will favor
selective acqui-hires and targeted M&A. They’re
buying execution, not experimentation.
Christian Lang
VP Data & AI and Western Europe IT
Stada Arzneimittel AG
„
Paul Ostwald
Co-Founder
morningcrunch
„
NEWSLETTER TIP
Expert Statements TREND 1
It’s time for impact. AI is entering its account-
ability era. We need to treat it like any other
strategic investment: tied to measurable out-
comes, scaled through core processes, and
built to last. Because sustainable impact beats
endless experimentation.
Walid Mehanna
Chief Data & AI Officer
Merck
„
While we saw many companies wanting to imple-
ment AI for the sake of technology (techno-
logy before use case), we’re now seeing a shift
toward companies starting with the use case
first and then mapping the right technology
to it; which can be AI, but doesn’t have to be.
Companies that solve real problems with AI will
stay. Companies that just want to “use AI” won’t
see any returns.
Maximilian Hahnenkamp
Co-Founder | Business
Scavenger AI
„
Expert Statements TREND 1
AI creates impact where it delivers tangible
value for customers or the core business –
often for both. Focusing on major priorities is
essential to move AI solutions from pilot to
production. At ING Germany, we design core
processes and connect GenAI and Agentic AI
with clear outcomes and scalable execution.
Johanna Biedinger
Tribe Lead Analytics & COO Transformation
ING Germany
„
For AI to deliver real value in 2026, companies
must define a clear strategic direction by
prioritizing the right use cases while actively
managing stakeholders and organizational
dynamics. Adoption and transformation will
not happen by accident; it will be the result of
intentional design and strong leadership.
Ibrahim Memis
Director Digitalization, AI & Cybersecurity
BEUMER Group GmbH & Co. KG
„
28
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
2026 marks the transition from selective AI pilot projects to robust opera-
ting models. The EU AI Act sets the framework for this: the general provisi-
ons and prohibitions have been in force since February 2, 2025, and rules
for general-purpose AI and the establishment of governance structures
will take effect on August 2, 2025.
On August 2, 2026, most of the obligations will finally take effect, including
transparency requirements and specifications for high-risk AI systems. For
high-risk systems that are used as a safety component of another product
according to the AI Act, the deadline will be extended by one year to August
2, 2027. At the same time, the EU is working on implementation aids and sim-
plifying adjustments without changing the basic logic. This raises the ques-
tion for companies: “How and for what purposes can we operate AI reliably
and safely?”
Clear rules are
shifting from man-
datory compliance
to a driver of returns
AI governance as a success factor
01 | COMPANY
TREND 2
© 2026 statworx & AI Hub Frankfurt
AI Trends Report 2026
29
AI Trends Report 2026statworx & AI Hub Frankfurt
2026 – 02 29
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
EU AI Act:
Four risk categories for AI systems
Unacceptable Risk
Prohibited
AI applications that violate
EU values by infringing on
fundamental rights
manipulation, social scoring,
biometric categorization,
emotion recognition
High Risk
Regulated
AI applications that may
negatively affect human
safety or fundamental rights
product components,
education, HR, creditworthiness
and lending, the legal system
STATUS RISK LEVEL STATUS
CATEGORY 1: COMPANY – TREND 2
AI applications with limited risk,
where transparency obligations
apply in order to build trust
chatbots, deepfakes, other
applications involving direct
interaction with humans
Limited Risk
Transparency requirements
AI applications that do not
pose additional risk and are
therefore subject only to
generally applicable law
spam filters, predictive
maintenance, video games, …
Minimal Risk
No additional requirements
Art. 5
Art. 6
Art. 52
30
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Slowing down to go faster
In this context, governance goes far beyond mere compliance with legal re-
quirements. While the AI Act sets the legal framework for risks, progressive
companies use governance structures to increase the efficiency of their
models. They translate principles into verifiable workflows with clear roles,
defined approvals, traceable decisions, and measurable results. Where this
framework is in place, speed and quality increase because discussions do
not start from scratch every time.
The effect is particularly evident where AI is used in regulated or reputation-
sensitive processes: human resources, lending, supply chains, or communi-
cations. In all these areas, well-designed controls enable greater speed be-
cause the risks are addressed: like in a car, where you brake before a curve
to navigate it safely and accelerate out of it.
Navigating a jungle of standards
For operational implementation, companies are currently looking for sup-
port in established standards, even if these do not fully reflect the AI Act.
ISO/IEC 42001 is the first auditable AI management system (AIMS) to establish
a logic for guidelines, risk management, and data quality that is compatible
with existing systems such as ISO 27001. The NIST AI Risk Management Fra-
mework, which originated in the US, supplements this with concrete working
practices, while the Generative AI Profile (NIST AI 600-1) provides guidelines
for generative models.
However, it is important to note that these frameworks are powerful tools,
but do not automatically guarantee compliance. They must be supplemen-
ted by specific requirements from the EU Commission and national authori-
ties such as the BSI or BaFin in order to be legally compliant. Nevertheless,
the AI Act defines the “what,” while standards such as ISO and NIST provide
initial structures for the “how.”
From a paper tiger to a real guardian
What does this look like in practice? Governance is effective when it is desig-
ned to cover the entire life cycle. The first step is to create a clear invento-
ry of use cases with risk classification: Which applications are non-critical,
and which are potentially high-risk? This is followed by impact assessments,
data and model profiles, human oversight design, and approval gates. In de-
velopment and operation, evaluation and monitoring pipelines take on the
role of “sensors”: tests before rollout (., factuality, bias), monitoring in
production (drift, failure patterns, prompt changes), and defined escalation
paths for unplanned incidents. This evidence feeds into audit trails and ma-
nagement reports, thus protecting against paper governance. NIST building
blocks and ISO/IEC 42001 requirements can be seamlessly combined here.
Across many large organizations, there is a clear belief that binding guide-
lines are a prerequisite for trust, both internally and externally. At the same
time, the implementation gap remains large: analyses by the World Econo-
mic Forum show that less than one percent of all organizations have fully
operationalized responsible AI practices to date. This creates a competitive
advantage for companies that consistently establish and expand their go-
vernance in 2026.
CATEGORY 1: COMPANY – TREND 2
31
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
When algorithms decide on jobs
The regulatory implications are particularly tangible when high-risk applica-
tions are used, for example in recruiting or creditworthiness checks for na-
tural persons, although not every application falls into the high-risk classi-
fication. For those systems that are affected, identical, strict requirements
will apply from August 2, 2026: human oversight, representative input data,
ongoing monitoring, and automatically generated logs. Logs must be kept
for at least six months, which means that emploxers need to take action.
Those who integrate these requirements into an AIMS at an early stage will
avoid duplicate entries later on and speed up approvals.
Beyond this high-risk class, however, governance also supports the spread
of AI in areas that are not subject to the strictest rules of the AI Act. In sup-
ply chains, for example, transparency and data quality are proving to be the
greatest levers for scalable AI processes. Regulation acts more as a cata-
lyst for efficiency here by preventing uncontrolled growth.
From intuition to impact: measuring what really matters
Organizational competence remains a critical issue. Without AI literacy in
specialist areas, compliance, and IT, processes remain ineffective. Recent
surveys reveal significant training gaps. This is a strong argument for es-
tablishing AI literacy as a governance metric and embedding it in approval
gates. After all, human oversight is only effective if employees clearly un-
derstand what they are authorized to approve and what they are required
to stop.
WHAT IS AI LITERACY?
AI Literacy refers to the ability to understand artificial intelligence,
use it competently, and critically evaluate its outputs. It encom-
passes a basic understanding of how AI systems work, the ability to
recognize their opportunities and risks, and the responsible, ref-
lective use of AI-powered applications in both everyday life and the
workplace.
Core building blocks of AI literacy:
• Foundational understanding of AI: Knowledge of how AI systems
function, where their strengths lie, and where their limitations
begin.
• Critical evaluation: The ability to question AI outputs, identify
bias, and understand uncertainty.
• Applied competence: Using AI tools effectively, efficiently, and
purposefully in day-to-day work.
• Ethics and responsibility: Awareness of the ethical, societal,
and legal implications of AI.
• Governance and compliance: Basic knowledge of rules, trans-
parency requirements, and responsible use (., the EU AI Act).
• Collaboration with AI: Understanding how human and artificial
intelligence can work together productively.
CATEGORY 1: COMPANY – TREND 2
32
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Whether something is truly effective from a technical standpoint depends
on whether it can automatically generate verifiable evidence. Documenta-
tion and versioning, standardized testing, and continuous monitoring with
alerting: these are the unspectacular but crucial ingredients of robust AI
operations. Governance is effective when it is present in tickets, pipelines,
and dashboards, not just in intranet PDFs.
For governance to become a lever for return on investment, a small but pre-
cise set of metrics is needed. Time-to-deploy and approval time (speed),
audit pass rate and incident rate (security), mean time to detect/resolve
for quality and drift events (resilience), and output quality and safety scores
in GenAI scenarios (effectiveness). In addition, metrics on third-party com-
pliance and AI literacy coverage provide insight into organizational maturity.
Studies suggest that transparency and governance accelerate scaling be-
cause they reduce coordination efforts and increase trust. A pragmatic
approach is recommended for implementation: first, set up the use case
inventory and classification, then fundamentally anchor risk management
frameworks, automate evaluation and monitoring pipelines, and finally eva-
luate the relevant KPIs on a quarterly basis and use them for continuous
control.
Standards as infrastructure for growth
The outlook is cautiously optimistic. The majority of AI Act obligations will
apply from August 2026, with supplementary aids and codes specifying the
implementation. Companies that now understand governance as an opera-
tional discipline are gaining speed and resilience. Not despite, but because
of clear rules. The lesson from the last waves of transformation applies here
too: standards and controls are not shackles. They are the infrastructure on
which true scaling is built.
CATEGORY 1: COMPANY – TREND 2
Expert Statements TREND 2
AI Governance has moved beyond mere box-
ticking; it now serves as a strategic control
system guiding how we deliver innovation
swiftly and securely. When embedded and
closely connected to real-world business
operations, it amplifies outcomes, accelerates
decisions, and safeguards long-term value.
As AI and generative AI adoption scales up,
organizations are increasingly recognizing that
AI governance is no longer just a compliance
requirement, but a strategic imperative.
Research shows that responsible AI practices
can drive competitive advantage by improving
product quality, attracting and retaining talent,
and opening new revenue streams.
Verena Dollberg
Program Director Corporate Strategy &
Digitalization | Fraport AG
„
Philippe Coution
Head of Digital Interaction & Automation (DIA)
Head of AI Quality (AIQ) | TÜV SÜD
„
Expert Statements TREND 2
In 2026, it’s no longer about who uses AI
first. What matters is who builds effective
frameworks for its responsible use. As long as
there is no consistent AI regulation, companies
must establish their own guardrails. In the end,
the companies that will lead the AI field are
those that use compliance and governance as
a competitive advantage.
The organizations that win in 2026 won’t be the
ones with the longest policy documents, but
the ones that translate rules into everyday
behavior: clear do’s and don’ts, role ownership,
and training that makes safe AI use feel easy.
Governance becomes ROI when it accelerates
adoption and prevents expensive “shadow AI”
risks.
Ingo Macht
Country Leader Deutschland
SAS
„
Rebecca Hundschell
Founder & CEO
The Female AI Club
„
Expert Statements TREND 2
Agentic AI challenges traditional accountability
structures, making an innovative and adap-
tive AI governance critical. Organizations that
embed a robust AI lifecycle governance into
their AI strategy, shift to a “safe uncertainty”
mindset, are more flexible to adopt new AI use
cases, to exploit their data for AI and are more
likely to achieve top-tier performance.
Catharina Glugla
Partner – Data, Cyber & Tech
A&O Shearman
„
36
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
The era of pure chatbots is over. 2026 marks the transition from genera-
tive to agentic AI. While builder tools radically simplify the creation of AI
agents, the real challenge is shifting dramatically: those who deploy auto-
nomous digital workers must manage them like human employees. With
clear rules, clean data, and strict governance.
The AI industry has fundamentally changed its direction. While 2024 was still
marked by amazement at text-generating models, big tech players such as
Microsoft, Google, and SAP are now going “all-in” on agentic AI. The differen-
ce is significant: a classic language model is reactive. It waits for a prompt
that gives it explicit work instructions. An agent, on the other hand, is proac-
tive and goal-oriented. It receives an abstract instruction (“Reorganize the
supply chain”), breaks it down into sub-steps, uses external tools (browser,
email, ERP systems), checks its results, and corrects itself as soon as it en-
counters obstacles.
However, in many places, this technological leap is encountering a working
reality that is not yet prepared for it. There is a large discrepancy between
technical feasibility and organizational maturity. The consulting firm Gartner
predicts that more than 40 percent of current agent projects will fail by 2027.
Not because of the performance of AI, but because of unclear processes,
lack of control, and lack of return on investment. 2026 will therefore be a
year of reckoning: it will separate organizations that view agents as a tech-
nological gimmick from those that can firmly integrate them into their value
creation as a scalable, digital workforce.
Where data and
processes are solid,
agentic AI can
reach its full
potential
Agents on probation
01 | COMPANY
TREND 3
© 2026 statworx & AI Hub Frankfurt
AI Trends Report 2026
AIResponds reactively to
inputs or predefined rules.
Acts proactively toward
goal achievement.
TRADITIONAL AI AGENTIC AI
37
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Click, click, chaos: when everyone becomes a developer
The hurdle to creating a powerful AI agent is lower than ever before. Plat-
forms such as Google‘s Vertex AI Agent Builder, IBM‘s watsonx Orchestrate,
and the German representative n8n, or tools such as OpenAI‘s Swarm Fra-
mework, have democratized development. What yesterday required weeks
of programming effort and a deep understanding of Python code can now
be developed in low-code or no-code environments, even by lay people. De-
partments can now independently define agents that take on specific tasks,
from summarizing complex compliance documents to preparing for sensiti-
ve customer appointments.
However, this new freedom carries a massive, often underestimated risk:
shadow AI. This refers in particular to the unknown use of AI, ., the use of
AI tools without IT approval. When marketing, HR, or sales departments build
agents on their own without adhering to central standards, the result is an
opaque proliferation of tools.
The problem goes far beyond inconsistent responses. An agent that has ac-
cess to the CRM system or internal APIs but does not have robust security
guidelines (guardrails) quickly becomes a risk. It could unintentionally send
confidential data to external parties or cause massive cloud costs through
endless loops. Without central orchestration, the IT department loses track
of which digital players are active in the corporate network and what goals
they should be pursuing.
Agents require supervision
But uncontrolled agent growth is not the only stumbling block. The success
of agentic AI is largely based on the data foundations that companies can
provide to their agents. S&P Global and IBM emphasize the need for “AI-ready
data.” An agent is only as good as the context in which it operates. If data
without clear metadata is stored unstructured in PDFs or silos, agents not
only hallucinate texts. In the worst case, they perform incorrect actions.
They post goods to the wrong cost centers or cancel valid orders. Data qua-
lity is not a nice-to-have, it is an operational necessity.
Furthermore, trust is a currency that must first be earned. Microsoft‘s AI
chief Mustafa Suleyman argues that autonomous agents only enjoy broad
trust in critical business processes when they achieve 99 percent accura-
cy. Currently, we are often only at around 80 percent in terms of technolo-
gy. Until this qualitative leap, which will probably only be fully achieved with
next-generation models, human supervision will remain indispensable for
the time being. Nevertheless, the trend is probably irreversible: it is expec-
ted that by 2028, 15 percent of daily work decisions will already be made
autonomously by AI.
Forward-thinking companies are responding to this by professionalizing AI
governance, for example through concepts such as the Agent Hub (Dataiku)
or Agent 365 (Microsoft). These centrally control which agent has which
CATEGORY 1: COMPANY – TREND 3
38
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
rights and how much budget it is allowed to use. At the same time, new
leadership roles are emerging, such as the Agent Boss, who delegates and
controls AI agents to increase the productivity of these digital workers. The
illusion that AI can simply be installed like traditional software and then for-
gotten is giving way to the realization that agents need to be managed like
interns: they require structured onboarding (training on company data), ri-
gorous testing in sandbox environments, and continuous monitoring of their
decision-making quality in live operation.
Where the digital workforce is already hard at work
Numerous pioneers who have already deeply integrated agents into their
processes demonstrate that the effort required for well-thought-out agent
governance is worthwhile. The applications go far beyond simple assistance
and are changing entire business models.
The telecommunications provider Verizon provides a particularly impressive
example. An AI agent based on Google‘s Gemini now supports around 28,000
service employees in customer contact. During ongoing conversations, the
agent accesses knowledge databases, suggests appropriate responses,
and assists with quotations, which significantly reduces processing times.
The decisive factor here is not only the increase in efficiency, but also the
added business value: According to Verizon, the use of AI assistance in the
service environment led to an increase in sales of around 40 percent wit-
hout any decline in customer satisfaction.
AI agents are also proving their worth in operational logistics, an area
characterized by high variance and unpredictability. DHL Supply Chain, for
example, is working with HappyRobot to deploy agents that independently
arrange appointments and check with drivers to see if everything is going
according to plan on their route. These are not simple chatbots, but sys-
tems that interpret unstructured calls or messages from drivers, compare
them with warehouse capacities, and make decisions in real time.
DHL Supply Chain is already successfully using HappyRobot’s AI agents across multiple regions.
(Image: DHL Group)
CATEGORY 1: COMPANY – TREND 3
39
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
In the field of highly skilled knowledge work, solutions such as OpenAI‘s Deep
Research Agent show how agents can independently perform complex mar-
ket analyses. A specially trained model independently identifies sources,
checks their reliability, aggregates data, and generates quotable reports.
Unlike a simple web search, the agent goes through several iteration loops,
refines its search strategy independently, and ultimately delivers a docu-
ment that saves analysts hours of painstaking work.
Agents are turning the economy upside down
The economic impact of this development will dominate the coming years.
Market analyses predict annual growth of over 45 percent for AI agents by
2030, reaching a market volume of around $50 billion. McKinsey expects
agents to contribute significantly to the $ trillion in value creation that
generative AI is expected to generate worldwide. This sum is not only due to
efficiency gains, but also to the ability to scale processes that were previ-
ously limited by human capacity constraints.
We are moving away from human-in-the-loop (where humans approve every
step of the AI) to human-on-the-loop (where humans only intervene in the
event of errors or escalations ). SAP is already talking about an “agent armada”
that collaboratively handles business processes from purchasing to ac-
counting to logistics. For companies, this means that those who get their
data processes under control now will be able to leverage these productivi-
ty advantages very soon and very quickly. Those who are still struggling with
data silos and unclear responsibilities, on the other hand, will fail due to the
complexity of agent control.
OpenAI: Introduction to Deep Research CLICK HERE!
CATEGORY 1: COMPANY – TREND 3
Expert Statements TREND 3
Shopping will become even easier: AI agents will
take over the entire purchasing process – from
finding products to making payments. Open
standards such as the Trusted Agent Protocol
and biometric authentication with passkeys will
provide the foundation for secure transactions
between autonomous agents.
In 2026, AI agents will be capable of orchestrating
end-to-end workflows, but material productivity
gains will accrue only where the foundational
layer is in place: high-quality data, well-defined
processes, deep domain expertise, and robust
governance. Banks should start with a clear
view on business value (. P&L impact) and
scale agentic AI only where these foundations
are ready.
Tobias Czekalla
Country Manager Germany
Visa
„
Nico Baum
Head of Solutions | Head of AI
Berenberg
„
Expert Statements TREND 3
Always-On-Research becomes essential for
active asset management in 2026 as information
on fundamentals and macroeconomic con-
ditions becomes available far faster than
monthly or quarterly releases – yet it only pays
off when agentic workflows run on clean data,
defined processes, and governance that turns
news flow into decisions while suppressing
noise from irrelevant information.
Agentic AI will redefine how organizations work.
AI agents will become embedded members of
insurance teams, acting autonomously within
clear mandates, collaborating seamlessly with
humans and other agents. This will create hybrid
teams that operate faster, learn continuously
and scale expertise – anchored by strong
governance and clear human responsibility.
Dieter Konrad
Expert Capital Market Data Science
Union Investment
„
Henry Byers
Head of Data & Advanced Analytics
Zurich Gruppe Deutschland
„
Expert Statements TREND 3
The sustainable use of AI and agentic AI
succeeds or fails based on the quality of
the underlying data. That’s why we focus
consistently on robust, accurate, and
governance-ready data and have developed
our own platform at EintrachtTech. It provides
the foundation needed to deploy AI agents
reliably, at scale, and with real business impact
to leverage AI’s full potential.
Agents, tools, and prompts are no longer
experiments, they are operational assets with
direct business impact. In 2026, quality, cost,
and risk must be measurable, comparable,
and continuously optimized. Without clear
operationalization, GenAI remains expensive
guesswork rather than scalable value creation.
Timm Jäger
CEO EintrachtTech
Eintracht Frankfurt
„
Daniel Schroter Thüm
Co-Founder
Promptic
„
Expert Statements TREND 3
Research is now turning into business, experi-
ments are becoming processes, and visions
are becoming reality – in 2026, AI must deliver on
its promises. AI is now good enough to improve
existing processes, but more importantly, to
create entirely new ones. And only when we
learn to think in AI processes will agents truly
work. We still have a lot to learn.
AI Agents aren’t magic – they’re a stress test
for your data and process foundations. Where
data is clean, processes are defined, and
governance is clear, agents can coordinate
end-to-end workflows and lift productivity;
where it isn’t, they multiply errors, costs, and
risk. Recommendation: build an “agent factory”
with approved platforms, testing, version
control, telemetry, and cost guardrails.
Andreas Wittke
Chief AI Officer
Technische Hochschule Lübeck
„
Dr. Marc Jäger
Lead Data Analytics & AI
BASF SE
„
44
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
2026 marks a turning point: Humanoid robots are leaving research labora-
tories and starting their shift in the real economy. Driven by new AI models
and industrial mass production, systems such as Digit and Figure are evol-
ving from expensive showcases to scalable colleagues. For companies,
the focus is now shifting from hardware issues to process integration.
For a long time, humanoid robots were considered science fiction or, at
best, an expensive gimmick for innovation departments. However, the com-
bination of a persistent shortage of skilled workers and technological ma-
turity has fundamentally changed the framework conditions by 2026. What
began with highly acclaimed pilot projects at BMW and in Amazon‘s logistics
centers is now transitioning into regular operation.
The robots are specifically filling gaps in logistics and manufacturing. They
are taking on tasks that are considered “dull, dirty, dangerous” and for which
it is difficult to find human personnel. A concrete example is Amazon‘s
“dead recycling” process: the monotonous collection and stacking of empty
transport crates. For humans, such activities often mean walking for miles
and repetitive physical strain.
The key difference to previous waves of automation is flexibility: these are
not fixed robot arms in cages, but mobile units that share infrastructure and
tools with humans. Anyone planning automation today would be wise to fac-
tor in humanoid colleagues.
Humanoid machines
are entering the
workplace
Roboter, ready for work
01 | COMPANY
TREND 4
© 2026 statworx & AI Hub Frankfurt
AI Trends Report 2026
45
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
A command such as “Clear away the empty container” no longer needs to
be translated into thousands of lines of code. The robot visually recognizes
the container, plans its route, and picks it up, even if the container is five
centimeters further to the left today than it was yesterday. This cognitive
flexibility allows for large-scale deployment in unstructured environments
(brownfield) for the first time, without the need to completely rebuild factory
floors for the machine.
Industrialization: The robot leaves the assembly line
With technological maturity comes economic scaling. In 2026, we will see
the first production facilities mass-producing humanoid robots, thereby
drastically reducing unit costs.
• USA: Agility Robotics operates RoboFab in Oregon, a 6,500-square-
meter (roughly the size of a soccer field) factory that is designed to
have a capacity of over 10,000 Digit robots per year. Figure AI has also
established highly automated production with its BotQ facility. To achie-
ve the desired production speed, Figure AI relies on processes from au-
tomotive engineering: instead of slow CNC milling processes, injection
molding and stamping are used, which drastically reduces the produc-
tion time of important components.
•
• China: Suppliers such as UBTech (Walker S series) are entering the mar-
ket with high volumes and already supply automotive manufacturers
such as BYD and Nio. The start-up AGIbot in Shanghai even caused a sen-
sation at the end of 2025 with a Guinness World Record: the longest dis-
tance ever traveled by a humanoid robot in one go. Agibot A2 covered a
distance of more than 106 kilometers in three days.
Figure Status Update
OpenAI Speech-to-Speech Reasoning
CLICK HERE!
The brain: Embodied AI changes everything
The real breakthrough that will make the use of humanoid robots econo-
mically viable in 2026 is taking place invisibly in the “brain” of the machines.
While classic industrial robots follow rigidly programmed coordinates, mo-
dern humanoids use so-called vision-language-action models (VLA).
Technologies such as those developed by Google DeepMind with Gemini Ro-
botics or OpenAI in cooperation with Figure AI transform hardware into em-
bodied agents. This means that the robot understands its environment se-
mantically and can draw logical conclusions. In demos, the humanoid Figure
01 has already impressively demonstrated what this means: in response to
the vague request “I‘m hungry,” the robot handed a person an apple. Not
because it was programmed to do so, but because it identified the apple as
the only edible object on the table.
CATEGORY 1: COMPANY – TREND 4
46
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
• Germany: Domestic industry is also positioning itself. The Metzingen-
based company Neura Robotics is working closely with automotive sup-
plier Schaeffler, which not only supplies parts for the robots but also
uses the humanoids in its own production. At the same time, the Mu-
nich-based company Agile Robots has taken over significant parts of
Thyssenkrupp Automation Engineering, thereby massively strengthe-
ning its industrial base.
Humanoid robots from Neura Robotics are set to be deployed at Schaeffler in the near future.
(Image: Schaeffler)
This industrialization is leading to astonishingly low unit prices. While early
prototypes often cost six-figure sums, today‘s models are reaching whole
new levels of affordability thanks to economies of scale. Tesla is aiming for
long-term prices of $20,000 to $30,000 for its Optimus robots. Basic models
such as the G1 from Chinese robotics pioneer Unitree start at around $16,000.
ely, some suppliers are positioning themselves even more affordably in the
entry-level segment, where models such as the Unitree R1 are already listed
for just under $5,900. This brings humanoids closer to the cost of a small to
mid-range car, and in some cases even significantly below it. This develop-
ment makes them attractive even for the ROI calculations of smaller com-
panies.
Renting instead of buying: robots as temporary workers
Despite falling hardware prices, many companies in 2026 are not opting for
traditional purchases, but rather rental models such as Robots-as-a-Servi-
ce (RaaS). Logistics giants such as GXO are leading the way: they rent robot
fleets for operational use, for example in a SPANX distribution center in At-
lanta. Companies do not pay for the metal, but for the hours worked, often
at rates of around $30. This hourly wage makes them directly competitive
with human labor. This approach shifts the risk from the investment balance
sheet to ongoing operating costs.
The challenge in 2026 will therefore lie less in financing and more in integ-
ration. For robots to become true colleagues, IT and security structures will
need to be adapted:
1. Infrastructure: Robots require stable Wi-Fi/5G in every corner of the
hall and intelligent fleet management. Platforms such as „Agility Arc“ or
solutions from Nvidia/Siemens are necessary to assign tasks and avoid
congestion in the warehouse aisle.
2. Safety: The collaboration between humans and humanoid machines also
requires new safety concepts (safety PLC, functional safety, emergency
stop systems), as traditional protective areas are to be eliminated.
CATEGORY 1: COMPANY – TREND 4
47
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
3. Process quality: The successful pilot run of Figure 02 at BMW demon-
strated the maturity of the technology. Over a period of ten months,
the robot installed complex car body parts with millimeter precision at
the Spartanburg plant (South Carolina). The lesson learned: The robot is
precise enough to replace a human employee in a highly complex, time-
critical task. However, this is only possible if the parts supply process is
equally reliable and standardized.
The year of proof
For many organizations, the use of humanoid robots is becoming a strategic
factor in their speed of innovation. 2026 is the year of revelation: the tech-
nology is here, the factories are delivering. Now, organizational adaptability
will determine who can realize the productivity gains.
Competition will continue to intensify, partly due to open-source approa-
ches from China that are further democratizing access to technology. Com-
panies should not wait until robots are perfect one day, but should make
their processes “robot-ready” now. Because two-legged robotic colleagues
are not just a thing of the future. They are already at the door.
The Figure 02 loads sheet metal parts into a welding machine at BMW’s production facility in
Spartanburg. (Image: Figure AI / Screenshot)
CATEGORY 1: COMPANY – TREND 4
Expert Statements TREND 4
Just like electricity once did, robotics will not
only transform a single industry, but every
industry – and not just industries, but our entire
way of life. Socially, culturally, economically:
robotics will reshape our world.
The keyword is operational integration. How do
security concepts, workflows, and interaction
with human specialists harmonize? The ability to
seamlessly adapt workflows will become the real
competitive advantage. Those who understand
in 2026 how to manage humanoid systems not
as foreign bodies but as an integrated part of
the workforce will transform pilot projects into
scalable productivity.
Kay Bärmann
CSO & Co-Founder
SYMBIOSIKA - EVERYDAY ROBOTICS
„
Thomas Kistner
Managing Director
Bricklog Deutschland GmbH &
„
Expert Statements TREND 4
Humanoid robots will, for the first time in 2026,
noticeably take over work in areas with labor
shortages or health risks rather than serving
as a show effect. Their contribution lies less in
replacement than in stabilizing processes with
high turnover. What matters is sober integration
into real processes instead of science fiction
expectations.
Stefan Dee
Process Excellence & AI Manager
Rosendahl Nextrom GmbH
„
There is a solid increase in real-world robotics
applications. What used to be research in the
past five years has now come much closer to
production. In my daily work, I have access to
robotics companies and their use cases years
before they go live. For startups and enterprise
companies alike, we have seen increased
investments and innovation in robotics. 2026 will
definitely see robotics growing.
Hans Ramsl
Principal Machine Learning Engineer
Weights & Biases by CoreWeave
„
50
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Artificial intelligence is about to move from the digital realm into the
physical world. Following years of experimentation, smart glasses and
hearables are expected to gain broader adoption by 2026. In the future,
these devices will no longer passively wait for input. They will evolve into
proactive, multimodal companions that see, hear, and understand. For
companies, this heralds the reality of the connected worker. Provided they
resolve the issues of compliance and data protection.
Until now, our interaction with AI has followed a rigid pattern: open the app,
type in a prompt, wait for a response. But this era is coming to an end. We are
heading toward a transition in which technology is no longer just a tool, but
becomes an extension of our own senses. AI is gaining eyes and ears. The
road ahead remains rocky, but instructive. The fact that early devices such
as the Humane AI Pin failed due to a lack of usability and ecosystems has not
slowed down the market, but rather cleaned it up. It is becoming apparent
that it is not the radically new gadget that will win, but the intelligent evolution
of what we already wear.
WHAT IS A CONNECTED WORKER?
A connected worker is a digitally enabled employee who is context-
ually supported by connected systems, data, and AI—allowing them
to work more productively, safely, and confidently.
Core characteristics of the connected worker:
• Real-time access to relevant information
• Digital assistance through AI and intelligent systems
• Seamless connectivity with processes, machines, and teams
• Continuous improvement driven by data and feedback
AI hardware turns
assistants into con-
stant companions
The second self
02 | WORK, ORGANIZATION &
COMMUNICATION
TREND 5
© 2026 statworx & AI Hub Frankfurt
AI Trends Report 2026
51
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
The Trojan Horse of wearable AI
The coming breakthrough of smart glasses – driven by models such as those
from Meta and Ray-Ban – is likely to be based on a simple psychological in-
sight: people don‘t want to wear computers on their faces, they want to look
good. The success of this device class will be based on the fact that they
are primarily fashion accessories and only secondarily tech gadgets. Sales
figures indicated early on that customers are buying the hardware even in
markets where the AI functions are not yet activated.
This hardware-first strategy acts as a door opener. Once the models are
powerful enough, the technology behind them will become virtually invisible.
Tech giants such as Alibaba and new alliances around OpenAI and design
legend Jony Ive are using this acceptance to weave AI deeply into our ever-
yday lives. The smartphone will remain in our pockets but will be used less
frequently over time.
CATEGORY 2: WORK, ORGANIZATION & COMMUNICATION – TREND 5
Meet the World‘s Most Advanced AI Glasses
Meta Ray-Ban Display @ Connect 2025
CLICK HERE!
Alibaba‘s first self-developed flagship
dual-display AI glasses — Quark AI Glasses
CLICK HERE!
52
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
When AI sees what you see
However, the real technological leap for 2026 lies not in smaller chips, but
in contextual understanding. An assistant that only processes text remains
blind to reality. The next generation of companions will literally see the world
through the eyes of their wearers.
Tourists standing in front of a historic ruin could have its history whispe-
red directly into their ear, triggered solely by their gaze, without any search
query. When you pick up a product in the supermarket, you get instant price
comparisons or warnings. Location, line of sight, and acoustic environment
converge in real time . The device will no longer act reactively, but anticipa-
tively: it “knows” what is relevant even before we ask for it.
Hands-free productivity for connected worker
This paradigm shift will not stop at the factory gates. What is convenient in
our private lives will become an efficiency booster in industry and the service
sector. AI wearables will enable true hands-free processes on a large scale
for the first time.
Maintenance technicians would no longer need to leaf through manuals;
the glasses would recognize the component, mark the defect in the field of
vision, and display the repair instructions. In international meetings, ear-
buds or smart glasses could act as simultaneous translators, breaking down
language barriers in real time. Documentation is done almost incidentally: dic-
tations by doctors or experts are automatically transcribed, structured, and
fed into the CRM system. Media breaks, which still cost time today, disappear.
The spy on your lapel: a compliance nightmare?
But this seamless integration comes at a price that organizations cannot ig-
nore. A device that is constantly listening and watching is potentially the per-
fect bug. When voice data is sent to the manufacturer‘s cloud as standard for
model improvement, as is often the case with consumer devices, this creates
a massive security risk. For companies, this means that hardware must not
be allowed to infiltrate offices unchecked. Clear no-wearable zones will be
needed in sensitive areas such as research and development. At the same
time, IT departments must review which enterprise solutions offer local data
processing or contractually guaranteed privacy standards. The naive use of
private smart glasses in the executive office could no longer be a trivial of-
fense in 2026, but a tangible compliance violation.
CATEGORY 2: WORK, ORGANIZATION & COMMUNICATION – TREND 5
53
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
From touch to thought-based interaction
Development will not stop there. While we are still getting used to voice con-
trol today, laboratories are already working on the next interface: neuronally
controlled wristbands that pick up signals directly from the wrist. This would
make interaction even more discreet : in the future, the mere thought of a
movement could be enough to trigger digital actions.
For now, however, the “second self” is coming our way. It will probably be a
pair of glasses, headphones, or a pin. Competition between AI models is now
being joined by competition between AI devices. It has the potential to make
us more knowledgeable in everyday life and faster at work, as long as we re-
tain control over who is listening in at the other end of the line.
CATEGORY 2: WORK, ORGANIZATION & COMMUNICATION – TREND 5
Expert Statements TREND 5
The second self will be useful – but it must
not become louder than the first: no AI
that overshadows situations or predefines
decisions. The breakthrough, however,
won’t come from features, but from trust –
the dealbreaker in sensitive areas. It takes
transparency, data sovereignty, consent,
control. Otherwise, “always-on” turns socially
toxic: cautious conversations, quiet rooms,
distrust.
Marcel Plaschke
Head of Strategy, Sales & Marketing
statworx
„
When it comes to AI wearables, the role of
design is often underestimated. Yet it largely
determines whether these products are
accepted or rejected in everyday life. Only when
wearables feel like natural accessories and the
technology fades into the background can AI
truly scale, build trust, and become a lasting
second self.
Camilla Jeck
Marketing Manager
statworx
„
55
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02statworx & AI Hub Frankfurt
The Internet of links is giving way to the Internet of answers. By 2026,
brands will no longer communicate directly with people, but with their
digital representatives. In this new consumer world, it is not clicks but AI
recommendations that determine market success. Those who are not
readable by machines will also remain invisible to humans.
For many, the classic search bar has become the epitome of the internet.
The digital gateway to the world. This era is coming to an end. Like vinyl re-
cords or VHS tapes, the search bar with its flashing cursor will soon become
a cultural relic: older people remember it nostalgically, younger people
hardly know what it actually was. In 2026, people searching for information
will no longer type keywords into a white bar and then click through ten blue
links. They will engage in a dialogue. Systems such as Google‘s AI Mode and
Amazon‘s shopping assistant Rufus are fundamentally changing the way we
shop online.
AI becomes the
command center
of marketing
The quiet intermediary
TREND 6
02 | WORK, ORGANIZATION &
COMMUNICATION
Google | Ask Search Anything CLICK HERE!
© 2026 statworx & AI Hub Frankfurt
56
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
The classic sales funnel of searching, clicking, visiting a website, and filling a
shopping cart is collapsing into a single, interactive interface. The result is a
radical shortening of the customer journey. For unprepared companies, this
means a historic loss of control: an AI chatbot is coming between the brand
and the customer, acting as a kind of gatekeeper for purchasing decisions.
The traditional B2C world is transforming into a B2AI2C (business-to-AI-to-
consumer) economy.
The central question for marketing decision-makers is no longer “How do we
generate traffic?” but
Beyond SEO: optimizing for answers
For years, search engine optimization (SEO) was the craft of convincing Goo-
gle and other search engines of one‘s own relevance. Optimization was based
on keywords, backlinks, and dwell time. But in a world of generative answers,
old SEO rules fall short. Keywords are losing importance, while semantic con-
texts are gaining ground. SEO is being replaced by GEO (generative engine
optimization).
Large language models (LLMs) do not search a database for hits; they calcula-
te the next logical word based on probabilities. Studies show that GEO measu-
res such as providing quotable statements, statistics, and clearly structured
facts can increase visibility in these AI responses by up to 40 percent.
At the heart of this change is what is known as entity salience. Modern AI
models do not simply make recommendations based on search terms. They
“How does the brand become part of the answer?”
evaluate the degree to which a brand is recognized as a distinct, verifiable
entity. If the AI cannot unambiguously identify brands, products, or experts ,
they will not appear in the response, no matter how high the keyword density
of the website is. This requires new metrics beyond page impressions. Tools
such as the Highwire AI Index make it possible for the first time to see how
often and in what context a brand appears in the responses of ChatGPT,
Gemini, or Claude.
Overview: SEO vs. GEO
CRITERION
Goal
Focus
Mode of
operation
Metrics
Structure
SEO
Visibility in
search engines
Keywords &
backlinks
Ranking
algorithms
Position & click-
through rate
Traditional
website
GEO
Visibility in AI-generated
answers
Trustworthy,
fact-based content
LLMs process and cite
content
Citation frequency
and order
Structured, LLM-ready
content (., )
CATEGORY 2: WORK, ORGANIZATION & COMMUNICATION – TREND 6
57
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Robot customers: the new super shopper
Perhaps the most radical change in the coming years goes one step further:
instead of comparing products themselves, customers send a personal bot
ahead. These customer bots take on the tedious work of searching, filtering,
and evaluating. Amazon is already delivering the prototype for this develop-
ment with Rufus. With remarkable results: customers who use Rufus while
shopping are 60 percent more likely to complete a purchase than customers
who do not use the assistant. Gartner predicts that by 2026, 20 percent of all
service interactions will be initiated by such machine customers. This com-
pletely turns the previous purchasing dynamic on its head.
A customer bot is not dazzled by colorful banners, storytelling, or influencers.
It makes purely rational decisions based on concrete data points. It checks
delivery times, CO₂ values, warranty conditions, and return rates and com-
pares them with thousands of competitors worldwide. If product data is not
structured and made available in a machine-readable format via APIs, the of-
fer does not exist for the bot. Marketing must learn to serve two target groups
in parallel: humans (emotional, visual) and machines (rational, data-driven).
For a long time, AI responses were considered neutral territory, a pure syn-
thesis of global knowledge. But in 2026, it became clear that they were beco-
ming a new, hotly contested advertising space. The large AI laboratories had
to cover their immense computing costs, and monetizing the “golden answer”
was the logical step. Analyses by Adthena have already confirmed the first
ads within Google‘s AI Overviews. The AI-based search platform Perplexity has
also begun to integrate native ads directly into generated texts, making them
almost indistinguishable from factual information.
At the same time, deals such as the one between Amazon and the media
companies Condé Nast and Hearst ensure that trustworthy editorial content
from magazines such as Cosmopolitan and Harper‘s Bazaar is exclusively in-
corporated into the recommendation logic of shopping bots. The open web is
thus noticeably losing importance. Visibility in AI is evolving into a pay-to-play
market. Those who do not pay or do not provide unique, non-replicable know-
ledge no longer appear.
CATEGORY 2: WORK, ORGANIZATION & COMMUNICATION – TREND 6
58
AI Trends Report 2026
© 2026 statworx & AI Hub Frankfurt
2026 – 02
Is this the end of customer relationships?
This change carries an enormous risk: the erosion of the brand. When trans-
actions take place directly in the chat interface, as the cooperation between
OpenAI and Etsy already demonstrates, customers lose all contact with the
manufacturer‘s brand world. They see no website, no corporate design, no
visual language. The brand experience is reduced to a text field and a trans-
action confirmation.
From product recommendation
to purchase: ChatGPT integrates
checkout in partnership with Etsy.
Learn more in Trend 20: “ChatGPT
goes Platform – OpenAI builds the
operating system of the AI era”.
Companies are at risk of becoming interchangeable white label suppliers for
the major AI platforms, relegated to logistics and production. In addition, a
new competition for the truth is emerging: Who defines which product the
AI recommends as “the best”? Since AI models tend to form plausible aver-
ages, innovative niche providers may find it more difficult to compete with
established market leaders.
But this upheaval does not have to be a threat. It also presents a tremen-
dous opportunity for quality providers and a renaissance of substance.
When customer bots compare products strictly on the basis of facts and
data, hollow advertising slogans lose their effect. Objective quality, excel-
lent service data, and genuine expertise are rewarded because they are
the only currencies the algorithm can process. Those who actually offer the
longest-lasting battery or the b
