AI, the Year of the Turning Point for Italian Companies: Adoption Doubled and First Real Impacts on Business

AI, the Year of the Turning Point for Italian Companies: Adoption Doubled and First Real Impacts on Business

In just twelve months, artificial intelligence has gone from buzzword to strategic infrastructure for a growing number of Italian companies. According to the latest Istat report “Enterprises and ICT – Year 2025”, 16.4% of companies with at least 10 employees use at least one AI technology, almost double the 8.2% in 2024 and more than three times the level in 2023 (5.0%).[2][1]

For a traditionally cautious production system like the Italian one, we are facing a change of phase: AI is no longer just experimentation by a few pioneers, but is becoming a cross-cutting competitive factor, with concrete impacts on marketing, operations, logistics, and finance.[1][2]

In this article, we analyze what the numbers really say, where investments are going, which adoption models are emerging, and how all this is reshaping business, between opportunities and risks.

The Explosion of Adoption: Not Just More Companies, But More Mature Use

The headline figure is clear: the use of AI in Italian companies has doubled in one year, going from 8.2% to 16.4% between 2024 and 2025.[2][1]

But the leap is not only quantitative.

More Technologies for the Same Company

Istat notes that it is not only the number of companies that declare using AI that is growing, but also the complexity of the projects:[2]

• the share of companies that use at least two AI technologies has increased from 5.2% in 2024 to 10.6% in 2025[2]
• combinations of intelligent automation, predictive analysis, and recommendation systems are spreading, often integrated with CRM, ERP, and existing data platforms[1]

This indicates that the most advanced companies are not limited to experimenting with a single generative AI tool, but are starting to build hybrid architectures, in which language models, machine learning algorithms, and automation systems coexist on real processes.[1]

Generative AI is Visible, But Not (Yet) the Majority

Generative AI dominates the public debate, but the overall picture is more articulated. An analysis conducted by Confindustria's Artificial Intelligence Sounding Board, with 241 use cases collected in 76 companies, shows that generative AI represents 18.3% of cases, while the majority of projects concern:[1]

• process automation (back-office, documentation, workflow)
• predictive analysis (demand, maintenance, risks)
• decision support systems and dynamic pricing

A picture emerges in which generative AI acts as a “visible” interface (copilot, chatbot, content generation), but the bulk of the economic value today is played on less flashy engines, integrated into operations, supply chain, and finance.[1]

Where AI is Really Changing Work: Sector Map

The Confindustria survey offers a detailed mapping of the sectors in which AI is producing structural impacts, going beyond superficial experiments.[1]

Healthcare and Life Sciences: Automation of «Invisible Work»

In healthcare, AI is used to reduce the burden of clinical bureaucracy and improve data quality:[1]

• automation of clinical documentation through voice recognition and language models: reports are transcribed in real time during visits, reducing errors and compilation times[1]
• advanced analysis of health data and images to support diagnoses, personalized treatment paths, and clinical trials
• use of digital twins and copilot systems to optimize pharmaceutical research and development, molecular simulations, and the study of complex biological networks[1]

The result is an initial shift of AI from an experimental research tool to an infrastructural lever for the efficiency of healthcare systems.

Smart Cities and PA: Sensors, Data, and Predictive Maintenance

In the perimeter of smart cities, AI is increasingly incorporated into critical infrastructures:[1]

• intelligent management of public lighting with sensors and predictive models for consumption and failures
• predictive planning of maintenance of urban greenery and infrastructures
• optimization of waste management and monitoring of large natural areas for prevention and rapid response to emergencies[1]

Here, AI becomes an engine of operational efficiency, with direct benefits on costs, sustainability, and service quality.

Finance, Purchasing, and Supply Chain: The Algorithm Enters the Margins

On the front of core business processes, AI is used to:

• optimize credit risk and customer solvency with advanced scoring models[1]
• predict the demand for raw materials and react in real time to price fluctuations and market offers[1]
• automate purchase management, reducing errors, cycle times, and operating costs

These are less visible use cases to the general public, but with direct impact on cash flow, margins, and resilience.

Logistics: 2025 as the «Year Zero» of Applied AI

In the world of logistics, industry observers speak of 2025 as the “year zero of AI”: the adoption of artificial intelligence technologies is transforming planning, tracking, and fleet management.[6]

According to recent data, the use of AI by Italian companies has doubled, but a strong gap between large and small companies remains, which risks translating into a structural gap in competitiveness in the supply chain.[6][2]

Dark Side of the Coin: Fragmented Adoption and «Unstructured» Use

Behind the positive numbers, Istat also photographs elements of fragility in adoption:[2]

• the share of companies that declare using AI without being able to indicate a specific business area is growing: from 15.5% in 2024 to 33.4% in 2025[2]
• the phenomenon mainly concerns small businesses, where adoption is often experimental, entrusted to individual departments or internal figures and not integrated into organizational processes[2]

In other words, a non-negligible part of the production fabric is in a phase of “surface AI”: tools used in an opportunistic way (to generate texts, slides, images, small automations) but without a real process transformation strategy.

The Obstacles: Skills, Regulation, Data, and Costs

Among the companies that still do not use AI, potential interest is growing: 11.5% have considered adoption, more than double compared to 2023.[2]

The main obstacles identified are:[2]

• lack of internal skills to select, integrate, and govern AI solutions
• regulatory uncertainty, especially in relation to the European AI Act and its interaction with GDPR, Digital Services Act, and other regulations[1][2]
• poor availability of adequate data (quality, governance, access) to feed effective models
• concerns related to privacy, security, and implementation costs[2]

The overall picture is that of an ecosystem in incomplete transition: AI is advancing rapidly, but its impact remains uneven between sectors, territories, and business size classes.[2][7]

New Regulatory Framework: The AI Act as a Game Changer for Companies

The acceleration of adoption is part of a European regulatory context in strong consolidation. The AI Act becomes the main pillar of an innovation governance that interacts with GDPR, Digital Services Act, and other key tools.[1]

Risk Classification and Role of the «Deployer»

The regulation adopts a logic of classification of AI systems based on risk, with increasing obligations as the criticality of the area of use increases.[1]

A central point for companies is the figure of the deployer:

• it is the company that uses a high-risk AI system in its process
• it has specific responsibilities on risk assessment, staff training, transparency, and continuous monitoring of the system[1]

For organizations that integrate AI into core processes (e.g. credit scoring, personnel selection, healthcare, critical infrastructures) this implies the need to build internal governance frameworks, with clear attribution of roles, audit processes, management of the life cycle of the models.

Impact on the Business

From Experimentation to Competitive Lever

The growth from 5% of companies in 2023 to 16.4% in 2025 indicates that AI is moving from a pilot project to a strategic asset in many realities.[2]

The most evident impacts concern:

• operational efficiency: automation of repetitive activities, reduction of errors, shortening of cycle times (back-office, documentation, customer service)[1][2]
• quality of decisions: thanks to predictive analysis, dynamic pricing models, optimization of risk and supply chain[1]
• personalization of the offer: recommendations, advanced segmentation, dynamic content for marketing and sales

The companies that started earlier to invest in data, cloud infrastructures, and analytical skills are now capitalizing on cumulative advantages difficult to recover for late adopters.

Marketing and Customer Experience: The Algorithm as a New “Creative Engine”

Although Istat data does not go into detail of the individual functions, the observation of Italian and international use cases indicates that digital marketing is among the first areas to benefit from AI:[1][8]

• assisted generation of content (texts, ads, creative variants)
• automated optimization of ADV campaigns on multiple channels
• predictive scoring of leads, churn, and propensity-to-buy
• fine segmentation based on behaviors and weak signals

In parallel, the proliferation of content generated by AI has created an information overload: according to a recent analysis, in 2025 poor quality content produced by AI has become mass, leading platforms and search engines to develop more sophisticated filters and ranking systems.[8]

For brands and agencies this means that AI is not only a production tool, but also target of new rules of quality and relevance: those who limit themselves to «pumping volume» risk being penalized by discovery algorithms.[8]

SMEs: Risk of Structural Gap

The strong presence of «unstructured» uses among small businesses highlights a risk of polarization:[2]

• on the one hand, large companies and medium-large groups that integrate AI in a deep way, with dedicated teams, clear governance, and scalability
• on the other hand, a vast audience of SMEs that experiments with generic tools without inserting them into a transformation strategy

If this gap is not filled, AI risks becoming a multiplier of competitive inequalities, rather than a widespread opportunity.[2][6]

Talents, Organization, and Work

The growth of adoption brings with it a direct impact on roles and skills:

• figures such as AI product owner, data steward, prompt specialist, AI compliance officer emerge
• the demand for hybrid profiles grows, capable of combining domain skills (e.g. operations, marketing, finance) with data literacy and familiarity with the models
• daily work is redesigned by copilots and intelligent assistants that support knowledge workers and operators

In parallel, the social perception of AI remains ambivalent: international surveys indicate that over half of workers fear that their job may be replaced by a machine in the coming years, confirming a climate in which adoption and replacement anxiety coexist.[3]

Strategies for a «Mature» Adoption of AI in the Company

From the data and use cases, some implicit guidelines emerge to move from tactical experimentation to long-term strategies.

1. Put Data at the Center

Without quality, governed, and accessible data, AI remains a laboratory exercise. The most advanced companies are:[1]

• consolidating unified data platforms
• defining clear roles of data ownership
• investing in data quality, lineage, and security

2. Connect Use Cases to Business KPIs

Winning projects start from clear economic objectives (cost reduction, revenue increase, NPS improvement, time reduction) and measure the contribution of AI in a structured way, avoiding the risk of the «permanent pilot».

3. Manage Compliance as an Enabler, Not Just a Constraint

The new European regulatory framework pushes companies to think about AI governance from the design phase:[1]

• audits on datasets
• traceability of automated decisions
• registers of high-risk AI systems
• impact assessment processes (in parallel with privacy ones)[1]

The organizations that integrate these elements into their procurement and development processes will be advantaged when the requirements of the AI Act become fully operational.

4. Invest in Widespread Skills, Not Only in «Centers of Excellence»

The lack of skills is today one of the main barriers.[2] The most mature cases show that AI becomes effective when it is not confined to a specialist team, but is understood and used consciously by managers, creatives, technicians, operators.

This requires continuous training programs, clear policies for internal use of generative AI, and support tools that make adoption safe and traceable.

A Scenario in Rapid Consolidation

2025 marks for Italy the transition from an elitist AI adoption to a phase of accelerated but still unequal diffusion. The increase in penetration, the growth in the complexity of projects, and the new European regulatory framework indicate that AI is no longer a marginal option, but a matter of strategic governance for companies and institutions.[1][2]

The trajectory of the coming years will depend on the ability of the country system – companies, supply chains, institutions, training – to transform this adoption race into a structural competitive advantage, and not into yet another patchy technological wave.