By Marco Mizzau
Artificial intelligence is not dematerializing the economy — it is shifting value toward energy, infrastructure and technical labor. As AI scales, the real constraint is no longer software, but power, grids and installation capacity. The result is a structural reallocation of capital, pricing power and geopolitical leverage.
AI is not destroying the physical economy. It is increasing the value of the infrastructure that makes it possible.
A superficial reading of current trends suggests that private equity will continue to pursue software, asset-light services, and high multiples, using AI as a further margin accelerator. A strategic reading suggests the opposite. We are entering a phase in which AI compresses value in standardizable cognitive labor, while reallocating pricing power, wages, and capital toward the physical infrastructure that enables AI itself.
This is not a simple sector rotation. It is a reclassification of economic value.
Exposure is highest in digital, administrative, software, finance, and formalizable knowledge work; much lower in physical, contextual, and manual labor. The chart in Anthropic’s report Labor Market Impacts of AI does not only measure “what AI can do,” but where theoretical capability is already close to observed use — and therefore where economic pressure is most immediate.
In the United States, Gen Z is revaluing technical trades such as electricians and HVAC technicians not out of industrial nostalgia, but for income, autonomy, and low substitutability.
An even more radical perspective emerges from Citrini Research’s macro work, The 2028 Global Intelligence Crisis, which suggests a potential super-cycle of investment in physical infrastructure — energy, power grids, manufacturing, and supply chains — as an indirect consequence of AI expansion.
This is where the paradigm shift for private equity lies.
For twenty years, the game has been to buy digital growth, finance high multiples, compress operating costs, and exit at higher valuations. The new game is different: acquiring physical scarcity, installation capacity, mission-critical services, and local networks that are difficult to replicate. Private equity will not move toward the physical economy because it is “more romantic.” It will move because defensibility, pricing power, and labor scarcity return there.
In the United States, this shift is already visible as a realignment between technology and the industrial base. The American model remains superior in rapid capital allocation, risk absorption, and financing new ecosystems. But AI has a dual effect: on one side, it strengthens leadership in computers, hyperscalers and software infrastructure; on the other, it devalues part of the white-collar professional class that supports consumption, credit and services. The U.S. economy is exposed because it is fundamentally a white-collar services economy. The mechanism is significant: if automation impacts high-income cognitive roles, the effect on aggregate demand could exceed the simple count of jobs lost.
China reads the same transition differently. It does not defend mid-level cognitive labor; it defends supply chains, energy, manufacturing, infrastructure, and industrial scale. In terms of power, China is better positioned than many Western economies if the new cycle favors hardware, electrification, supply chains and installation capacity. Its vulnerability is not industrial but geopolitical: export dependence, technological containment, bottlenecks in advanced semiconductors and a global financial system still denominated in dollars. For this reason, the bifurcation between cognitive AI and the hard-asset economy tends to favor those who control end to-end physical capacity.
Russia remains less relevant in the advanced AI layer but continues to be central in a world where energy, defense, commodities, and supply chain security regain importance. It does not win at the technological frontier; it influences the systemic cost of the transition. In a cycle dominated by hard assets, its ability to disrupt energy, logistics and European security restores relative weight.
Israel occupies a distinctive position. It is too small to dominate infrastructure scale, but highly effective at converting technology into security, dual-use capabilities, and intelligent defense systems. In an environment where capital seeks exposure to defense tech, sensors, military automation, and mission-critical systems, Israel remains one of the most efficient models for translating innovation into strategic capability. It is not the center of the global physical economy, but it is a multiplier of value in high-tech, security-intensive segments.
The European Union is the most fragile case. It has regulatory demand, an industrial base, manufacturing know-how, and an objective need to invest in networks, defense, energy, and automation. But it suffers from slow decision-making, high energy costs, fragmented capital markets, and an inability to scale at American speed or Chinese discipline. Mario Draghi’s recent analysis is explicit: Europe faces a competitiveness gap, particularly in energy and in financing transformation. This means that the “AI + infrastructure” thesis is valid in Europe as well, but requires greater public-private coordination and more patient capital.
The decisive point is not that AI replaces labor. The decisive point is which labor becomes abundant and which becomes scarce. As highlighted by Anthropic, AI coverage — both theoretical and observed — is highest in linguistic and digital occupations such as software, administration, business, finance and much lower in contextual physical work such as construction, repair, and production.
AI does not impact “labor” in the abstract. It impacts labor that can be converted into text, rules, and standardized sequences.
From this emerges the new chokepoint. If value shifts away from standardizable knowledge work, it concentrates in four layers of control: compute, energy, power grids, and installation capacity. AI does not exist in an abstract cloud. It exists in data centers, chips, turbines, transformers, cooling systems, interconnections and the technicians who install and maintain them.
The International Energy Agency estimates that data center electricity consumption could nearly double by 2030, reaching around 945 TWh in a base case scenario, with annual growth of approximately 15% — far exceeding overall electricity demand growth. This is the link between Anthropic’s micro-level findings and Citrini’s macro thesis: every unit of cognitive automation increases dependence on the physical economy.
The distinction is no longer between tech and non-tech. It is between builders, controllers, and dependents.
Builders construct and install physical capacity: electrical contractors, HVAC services, utility infrastructure, system integrators, and civil-industrial services. Controllers manage high-rent bottlenecks: chips, grid nodes, power equipment, industrial software and defense systems. Dependents are those selling easily disaggregated cognitive labor: standardized professional services, digital outsourcing, operational media and document-based middle management.
Citrini suggests that the “AI infrastructure complex” — data centers, energy, semiconductors, and networks — may continue to grow even in a slowdown of cognitive-labor-driven demand. The insight is structurally correct: the winners are not those who “use AI in PowerPoint,” but those who own the material constraints of its deployment.
From a capital flow perspective, the shift is clear. Traditional software-centric private equity relied on recurring revenues, margin expansion through efficiency and valuation multiples sustained by growth scarcity. If AI reduces the marginal cost of many cognitive functions and lowers barriers to replicating mid-tier software products, then part of software is no longer scarce — it becomes contestable.
Citrini highlights this risk by pointing to downgrades in PE-backed software debt and defaults linked to structural AI disruption. While the tone is deliberately extreme, the vector is realistic: not all software is obsolete, but software lacking deep integration, proprietary data, or embedded workflows may experience significant multiple compression.
Capital therefore begins to favor what cannot be “prompted away.” Private equity is not only seeking growth. It is seeking predictability of cash flows, market fragmentation, local pricing power, low substitutability and consolidation potential.
Electrical contractors and HVAC services satisfy nearly all of these conditions: structural demand, fragmented markets, aging ownership, certification requirements, strong local presence and limited automation potential. U.S. Bureau of Labor Statistics data confirms that electricians and HVAC technicians are growing faster than average, with many openings driven by retirements and generational turnover.
Capital will follow three trajectories:
First, selective refinancing: less tolerance for weak-moat digital assets, greater appetite for industrial services with recurring contracts and visible capex.
Second, compression of the valuation premium between “tech” and “non-tech”: physical infrastructure, utility adjacencies and industrial services will be re-rated as enabling layers of AI growth.
Third, labor and wage reallocation: where AI creates cognitive abundance, capital captures margins; where it creates dependence on physical capacity, skilled labor regains bargaining power.
The case of a 23-year-old U.S. electrician building income, a business, and even a media presence illustrates that technical work is not returning as simple blue-collar employment. It is returning as local micro-capitalism.
The most interesting blue-collar profile of the next cycle is not the generic wage worker, but the licensed, certified technician with excess demand, direct client relationships and the ability to scale into a small platform. For lower middle-market private equity, this is natural terrain.
For a strategic investor, winners will not be “all AI companies.” They will be assets that monetize the scarcity created by AI.
First: electrical infrastructure, utility services, grid modernization, transformers, switchgear, interconnection and engineering-installation capacity.
Second: HVAC, cooling, thermal management, and building electrification, especially where AI drives data center demand and energy transition drives retrofits.
Third: defense manufacturing and defense tech, as geopolitical pressure and rising NATO spending reopen a multi-year cycle of orders and capacity.
Fourth: industrial automation integrators — not necessarily pure robot makers, but those deploying real automation in factories and plants.
Relative losers are business models based on low-moat cognitive intermediation: standardized consulting, marketing agencies, legal process work, document outsourcing, repetitive media content, parts of horizontal SaaS and professional services selling formalizable labor rather than rare judgment.
Anthropic does not claim these sectors will disappear immediately. It shows they have the highest theoretical and increasing observed exposure. For capital, this is sufficient: when margin compression risk rises, multiples adjust before income statements do.
In the short term (24–36 months), the most robust strategy is a barbell: exposure to AI bottleneck controllers (energy, data centers, power equipment, cooling, selected semiconductors, industrial software) alongside exposure to fragmented local services in the physical economy, where private equity can execute disciplined roll-ups.
In the medium term (3–7 years), success will depend on converting these roll-ups into systems of procurement, training, dispatching, and cross-selling. In the long term (7–15 years), winners will be those building networks of hard-to-replace skills.
The real asset will be orchestrated skilled labor, not replaceable administrative labor.
The second-order effect is decisive: more AI means more data centers; more data centers mean more electricity, connections, cooling, permitting, maintenance, inspections, and technicians. The third-order effect is even more interesting: scarcity of technical labor creates space for training platforms, staffing systems, vertical operational software and specialized labor marketplaces.
This is an underappreciated frontier: not just investing in electricians, but in the organizational infrastructure that multiplies their productivity.
For Europe, and Italy in particular, this shift is both a threat and an opportunity. The threat is clear: Europe is weaker in global-scale AI champions, faces higher energy costs and suffers from slow decision-making. The Draghi report explicitly identifies both the competitiveness gap in energy and the need for massive investment.
The opportunity is less discussed. If value shifts toward grids, power systems, industrial automation, defense, maintenance, interconnections, and hard infrastructure, Europe regains relevance in real assets.
Italy has more of these assets than it appears. Terna plans over €23 billion in grid investments between 2025 and 2034. This is not just a utility story. It is the foundation for data centers, industrial electrification, selective reshoring and high-value technical services. Italy has also three underestimated assets: Mediterranean positioning, a base of technical SMEs and distributed technical human capital.
Its limitation is not demand, but the ability to aggregate capital, managerial capabilities, and scale vision. Italy has many high-quality PE targets, but few national consolidators.
This is where European private equity may find one of its few true advantages: acquiring fragmented technical capacity before it is repriced.
The core thesis is simple. AI is not making the physical economy irrelevant. It is re-evaluating it.
Anthropic shows where cognitive labor becomes exposed. Citrini identifies the macro mechanism. When intelligence becomes cheap, value migrates toward energy, infrastructure, installation capacity and scarce technical labor.
Author: Marco Mizzau – Strategic analyst focused on geopolitical economy, artificial intelligence and global power dynamics. He has held senior executive roles in international companies and serves as Chairman of Blacktrace. His work examines how technological systems, energy infrastructure and capital flows reshape the competitive position of states across the United States, China, Russia, Israel, and Europe. He advises U.S. investment funds on private equity strategy and capital allocation across energy, infrastructure and industrial systems.
(The opinions expressed in this article belong only to the author and do not necessarily reflect the views of World Geostrategic Insights).