Summary of Key Points
Recently, there has been an interesting shift in the capital markets: for over a decade, capital has favored "smart businesses" such as software and platforms (which have zero marginal costs and are easy to replicate). However, now it is flowing towards heavy-asset manufacturing industries and family-owned businesses in regions like Italy—these companies are referred to by private equity as HALO (Heavy Assets, Low Obsolescence Risk). The reasoning behind this is that AI has made "answers" (such as solutions, code, or posters) cheap and abundant, but the "result systems" that can turn these answers into actual outcomes (orders, deliveries, cash flows)—i.e., factories, equipment, customer relationships, data chains, and responsibility frameworks—are becoming scarce. Capital is not abandoning AI; instead, it is looking for the "complementary assets" that allow AI to be effectively implemented. This also highlights that "heavy-asset businesses," while valuable, are not omnipotent and need to be transformed with AI to maintain their value.
1. AI is widely adopted, but profits haven't increased—installing the engine is easy; getting the car running is difficult
Almost every company is adopting AI (88% of organizations have used it, such as in customer service robots and sales email tools), but only 39% say that AI has impacted their profits. Why? For example, creating a chatbot for customers is simple, but to truly make money with AI, you need to change data access rights, approval processes, and define job responsibilities. These changes can stir up old interests and expose organizational issues, so companies prefer to conduct "pilots" rather than completely restructure their operations. As a result, while AI saves time, it hasn't led to higher revenues or fewer employees.
2. AI makes answers cheap; only those that can be implemented are valuable
AI can generate multiple market plans, code, or posters in minutes, but these answers themselves are of little value. What matters is the process of turning them into actual results. For instance, AI can write a perfect sales email, but it doesn't have your customer list or a history of trust; it can suggest energy-saving factory improvements, but it can't replace equipment or pass regulatory inspections; it can create autonomous driving code, but it can't handle unexpected issues like sensors being covered in mud or pedestrians suddenly appearing on the road. These implementation steps (customer consent, equipment modification, responsibility tracking) together form a "result system." AI is good at creating things that seem right, but the business world buys what is proven to be useful—this difference is where future profits lie.
3. Capital hasn't abandoned AI; it's looking for its "other half"—heavy-asset companies are becoming sought after
Italy has become a favorite among private equity investors: private equity transactions in Italy increased by 16% in 2025, and global industrial private equity transactions reached a six-year high in the first five months of 2026, while countries like the UK, France, and Germany are seeing slower growth. Why does capital prefer Italy? Because it has many family-owned manufacturing companies with low debt, decades-old equipment, and stable global customers—these are all complementary assets for AI implementation. The advantage of software companies is scalability (one product can be sold to ten thousand customers), but AI shortens the lifespan of software functions (customers may develop their own solutions). In contrast, real-world assets like satellite optical systems, train fire protection systems, and mining maintenance teams cannot be easily replicated. Capital buys these companies not to abandon AI but to acquire its "other half," enabling AI to create actual value.
4. The result system is the true moat—don't idealize heavy-asset businesses too much
What is a "result system?" For example, SEMA in Italy's train automatic fire protection system: the key isn't just the increased accuracy from 97% to 98%, but how many false alarms were reduced and how many hours of downtime were avoided, as well as whether customers are willing to renew the contract. Companies with result systems can turn AI solutions into revenue and adjust their systems based on outcomes (e.g., by upgrading equipment or modifying processes). This is a hard-to-replicate advantage.
In conclusion
AI has made answers increasingly cheap, but the next wave of scarcity lies in who can turn those answers into actual results. Whether it's companies or individuals, those with "result systems" will have the power to set prices.