According to NYU Stern data from January 2026, revenue per employee reaches 1.758 million dollars in entertainment software, compared to 28,067 dollars in general electronics, and not in consumer electronics. This gap describes very different aggregate revenues per employee depending on the listed categories retained; it is not sufficient to establish macroeconomic causality linked to marginal costs. The question that opens up is not whether this dynamic is just or unjust, but to understand what drives it, what it implies for workers, and how businesses and public authorities can broaden its benefits.
The Essentials
- Revenue per employee reaches 1.76 million dollars in entertainment software versus 28,000 dollars in consumer electronics, according to Vena Solutions 2026: a gap that reflects software’s ability to reproduce at nearly zero marginal cost.
- The central mechanism is that of increasing returns: a video game sold a million times does not cost a million times more to produce, whereas assembling a million smartphones mobilizes a million units of labor and material.
- AI amplifies this movement by automating repetitive tasks in physical production, compressing workforce numbers without necessarily increasing the value generated by each remaining position.
- This concentration of value in software creates mounting pressure on training and career trajectories: the most productive jobs require skills that traditional education has not yet supplied at the required scale.
- Public policies that ignore this bifurcation risk leaving a growing fraction of the workforce in sectors where wage compression is strongest.
Sixty to One: What This Gap Really Says
The figure deserves to be broken down before being interpreted. Revenue per employee is not a salary: it is turnover divided by the number of employees. In entertainment software, it incorporates revenues from subscriptions, licenses, and digital sales that accumulate across global user bases, without the production cost of an additional copy being significant. In consumer electronics, each unit sold mobilizes components, energy, transport, and assembly labor. The ratio therefore reflects two fundamentally different economic logics, not two levels of talent or effort.
This distinction has a name in economics: increasing returns to scale. A studio like Epic Games or Valve can multiply its revenues by ten without multiplying its workforce by ten. A television or headphone manufacturer, meanwhile, remains subject to a linear constraint: more products sold means more variable costs. The physical model plateaus where the digital model takes off.
Economists like Philippe Aghion studied, beginning in the 1990s, the role of innovation and intellectual property in Schumpeterian growth. Global digital distribution participates in this phenomenon, whose visibility in corporate data has strengthened.
The Geography of Value in the American Economy
The sectoral ranking from Vena Solutions places this gap within a broader hierarchy. The ranking places entertainment software, real estate development, and brokerage/investment banking at the top. In the table, general electronics sits near the bottom of the sample, but the food industry and consumer and office electronics are not; no textile line is published. Between the two, an intermediate zone where production and software coexist, such as automobiles or healthcare.
This geography is not fixed, but it is structured by durable forces. The first is the zero marginal cost of digital reproduction. The second is protection by intellectual property rights, which allow software publishers to maintain high margins without price competition. The third is the network effect: an online gaming platform gains in value as its user base grows, creating barriers to entry that physical production does not know.
For the United States, this geography has a direct geopolitical dimension. The battle for semiconductors illustrates this tension well: the country seeks to recapture physical production capacity in a strategic sector, while front-end design and manufacturing are both essential, high-value-added, and technology-intensive activities. Intel, TSMC Arizona, and Samsung Austin manufacture; Nvidia and Qualcomm are primarily chip designers calling upon foundries, while Apple designs, markets, and has its products manufactured through a global chain of suppliers. The gaps between these activities depend on the segments of the value chain considered.
AI Reshuffles the Cards, But Not Equally
Artificial intelligence fits into this dynamic in an ambivalent way. In software sectors, AI can improve certain productivity indicators in certain tasks, but observed effects vary by tool, task, and experience; generalization to graphic designers and content teams is not established here. Revenue per employee in these sectors can evolve with changes in production, workforce, and revenues.
In physical sectors, AI operates differently. It optimizes supply chains, reduces production waste, improves defect detection in manufacturing. But it rarely compresses the cost of raw materials or transport, which remain the dominant line items. Automation can modify or reduce certain production and inspection jobs, but BLS projections also associate it with production and productivity gains; the specific effect of AI is not isolated.
This asymmetry deserves attention. In manufacturing, automation is projected to increase production while slightly reducing employment; one cannot attribute the additional ratio primarily to workforce reduction, nor impute it to AI alone. For workers affected, the difference is considerable.
Strategies of Physical Companies to Avoid Remaining Constrained
Saying that physical production is structurally disadvantaged does not mean it is static. Several strategies allow industrial companies to move up the value hierarchy, and some succeed at this.
The first is software integration. Apple is the canonical example: the company manufactures little itself, but controls the software ecosystem surrounding its devices. The App Store, iOS, and subscription services now represent a growing share of its revenues, with margins that are more software than hardware. Nike followed a similar trajectory with its coaching applications and digital subscriptions. In both cases, the physical company succeeded in capturing a layer of software value in addition to its products.
The second strategy is moving upmarket toward customization and service. A machine-tool manufacturer that sells maintenance contracts based on IoT sensors and fault-detection algorithms now sells as much software as metal. Siemens, Schneider Electric, and Rockwell Automation have largely pursued this path, transforming their industrial divisions into integrated solution providers where software margins compensate for hardware compression.
The third is strategic relocation to high-value segments of the chain. Rather than producing everything, some American industrialists abandon mass assembly to low-cost zones and concentrate their workforce on design, prototyping, and precision engineering—segments where revenue per employee rises significantly. American industrial policy since the CHIPS Act aims in part at this result: recreating on American soil the links with higher value added.
The Pressure on Skills and Training
The sectoral bifurcation revealed by this data has a direct implication for career trajectories. High revenue-per-employee jobs in software require skills in development, experience design, digital product management, and increasingly, AI engineering. These skills are not acquired in traditional training pathways that supply the physical industry.
In the United States, this mismatch has been documented for several years. The Bureau of Labor Statistics projects strong growth in software development jobs for the 2022-2032 decade, while assembly and industrial operation jobs stagnate or decline in volume. The difficulty is not the absence of software jobs: it is access to training that leads to them, often lengthy, costly, and concentrated in a few metropolises. Training for AI jobs remains insufficient to meet needs, and their development is one of the most urgent open projects to prevent the concentration of value in software from being compounded by geographic and social concentration.
A few players are trying to broaden access. Development bootcamps like General Assembly and Lambda School have promoted the promise of rapid career transition. Major platforms like Google, Microsoft, and Amazon have launched certification programs in cloud computing and AI, often free or low-cost, explicitly targeting workers in transition. Results are encouraging but limited in scale: training tens of thousands of people does not solve a tension affecting millions. Europe is grappling with the same issues and seeking to build training systems adapted to AI without yet finding a universally reproducible model.
Concentration, Access, and Public Policy
The productivity gap between software and physical production raises a question of economic policy that Vena Solutions data makes difficult to sidestep. The concentration of value in certain sectors, companies, or geographic zones can have different effects on tax revenues, skilled employment, and social mobility.
Several levers exist to correct this dynamic without hindering it. The first is tax policy: in the United States, long-term capital gains and qualified dividends are generally taxed at preferential rates compared to wages; intellectual property royalties and business income do not automatically benefit from the same treatment. A review of this asymmetry would not change fundamental economic returns, but would redistribute some gains. The debate on taxation of tech giants, active in both Washington and Brussels, falls within this logic.
The second lever is public investment policy in R&D and digital infrastructure in disadvantaged territories. American federal programs like the Regional Technology and Innovation Hub, launched as part of the CHIPS and Science Act, seek precisely to decentralize software innovation hubs outside the coasts, toward Midwest and Southern states most affected by deindustrialization over the past thirty years.
The third lever is competition regulation in software sectors. Philippon and Gutiérrez have argued that from the late 1990s onward, concentration and profits have progressed more in the United States than in Europe in many markets; telecommunications is an example, but generalization to several technology sectors requires more precise definition. Excessive concentration in software and platforms reduces incentives for innovation, compresses wages for developers outside large companies, and limits the number of new entrants capable of creating value and jobs.
These policies will not by themselves modify the revenue-per-employee gaps observed between sectoral categories. They can, however, broaden access to high-productivity sectors, redistribute some gains to workers and territories not yet participating in them, and prevent the sectoral bifurcation from crystallizing into permanent fracture.
The challenge for years to come is the speed at which training tools, investment policies, and competition regulation can keep pace with an acceleration in technological dynamics that is not slowing down.
Sources
- Vena Solutions – Average Revenue Per Employee by Industry 2026
- Bureau of Labor Statistics – Occupational Outlook Handbook, Software Developers (bls.gov/ooh/computer-and-information-technology/software-developers.htm)
- Thomas Philippon – The Great Reversal: How America Gave Up on Free Markets, Harvard University Press, 2019
- Philippe Aghion, Peter Howitt – The Economics of Growth, MIT Press, 2009
- CHIPS and Science Act – Regional Technology and Innovation Hub Program (EDA, eda.gov)
