Italy bets on physical AI to turn its robotics edge into a productivity revival

Industry leaders, manufacturers and start-ups are positioning Italy for the next industrial AI race, but adoption among SMEs still lags behind Europe and remains focused mainly on efficiency rather than innovation.

Italy’s manufacturing debate is increasingly centring on what industry leaders are calling “physical AI”: systems that do not stop at software predictions but can sense, decide and act in the physical world. At recent annual meetings in Milan and Turin, the heads of Assolombarda and the Turin industrial association both framed it as the next industrial frontier, and as a potential lever for productivity in a country whose growth problem is increasingly tied to output per worker rather than headline investment alone. According to the account published by Agenzia Digitale, the argument is that Europe’s opportunity lies in bringing artificial intelligence into real production environments, not only into office software and digital services.

That message has a clear industrial logic. In a separate report carried by Borsa Italiana, Assolombarda president Alvise Biffi said AI should help manufacturing firms create value, and argued that many small and medium-sized businesses have the products and technical skills they need but not always the know-how or scale to extract full benefit from new technologies. He also set an explicit target of lifting productivity in industrial SMEs by at least 10%, which he said could translate into an additional €2.4 billion in value for the region.

The regional picture is stronger than the national one, but both point in the same direction. Agenzia Digitale says an Assolombarda survey with the Politecnico di Milano found that 49% of the firms sampled across Milan, Monza Brianza, Lodi and Pavia already use AI, with another 18% planning to adopt it soon. The same piece notes that adoption is especially high among large firms and start-ups, but is also spreading among micro, small and medium-sized businesses. Even so, the report warns that usage remains concentrated in efficiency gains, process simplification and lower operating costs, while only a minority are using AI to design new products or services.

At national level, the picture is less encouraging. Agenzia Digitale cites the latest I-Com report as saying that Italy’s SMEs have made progress in basic digitalisation, but that AI adoption is still lagging behind the European average. The report also highlights a widening gap between SMEs and larger companies, and says that, if current trends continue, Italy would fall well short of the EU’s longer-term ambitions. The same analysis underlines that manufacturing remains one of the slower sectors for AI uptake, both in Europe and in Italy.

That matters because manufacturing is precisely where physical AI could be most valuable. In a country with a deep industrial base, a strong robotics tradition and a dense network of suppliers, the technology could support automation in complex environments, customised production and on-site decision-making. Agenzia Digitale notes that Italy is the second-largest robotics market in Europe and that Turin is one of the continent’s leading robotics centres, giving the country a stronger starting point than many of its peers. The same article points to industrial data, edge computing and machine-level intelligence as natural building blocks for the next phase.

The commercial momentum is already visible. Agenzia Digitale reports that recent academic and market studies show a sharp rise in attention to physical AI since 2025, with scientific publishing and internet interest accelerating quickly. It also cites a report from Mitsui & Co. Global Strategic Studies Institute saying venture capital investment in physical AI and robotics reached $26 billion in 2025, up more than 80% from the previous year, and is expected to climb further this year. The implication is that what was recently a niche topic has become a major global race.

Italian companies are trying to position themselves inside that race. Comau, the Turin-based industrial automation group now backed by a private equity owner but still partly tied to Stellantis, has deep roots in automotive robotics and system integration. Its recent partnership with Nvidia, reported by the company, is designed to bring advanced simulation and accelerated computing into future robot controllers and industrial automation systems. In practice, that could help Comau move from traditional robotics towards more adaptive, AI-enabled machines, including applications that rely on edge computing.

Generative Bionics is taking a different route. The Genova-based start-up, spun out of the Istituto Italiano di Tecnologia, has built its business around humanoid robotics and physical AI data infrastructure. According to the company’s announcement, it has raised €70 million with backing that includes CDP Venture Capital, AMD Ventures, Duferco, Eni Next, RoboIT and Tether, and plans to create three physical AI data factories, beginning with a 5,000 square metre site in Italy due to come on stream by the fourth quarter of 2027. The first facility would support robot training at scale, while later sites are intended to serve Europe and global markets.

The policy debate is now catching up with the industrial one. Agenzia Digitale says Industry Minister Adolfo Urso has signalled that the government wants both to encourage advanced robotics production and to monitor its impact on jobs. That is likely to remain a delicate balance, especially in a period of tight public finances and an exhausted post-pandemic recovery programme. A more practical question is whether Italy can use targeted incentives, industrial procurement and a stronger domestic supply chain to support adoption without trying to pre-select winners too early. For the moment, the most credible path appears to be a mixed model: encourage investment, strengthen the national and European framework, and let the market decide which companies turn physical AI into lasting industrial advantage.