Tech • AI • Robotics • Game
Where the value of AI really goes (and 6 ways to capture a share)
8/10TL;DR
Artificial intelligence is making expert services dramatically cheaper for consumers while concentrating most long-term gains in capital, infrastructure, and platform ownership rather than wages.
KEY POINTS
Consumers gain, workers do not
AI has created a large consumer surplus: services resembling legal advice, tutoring, coding help, writing support, and basic medical guidance are available around the clock for roughly €20 a month, when similar expertise once cost thousands in hourly fees. That benefit is real, but it does not show up cleanly in GDP, making headline growth look weaker than the lived improvement in cheap digital services.
Private markets captured the upside early
The largest gains have accrued to labs and their backers, including cloud giants, venture capital, and some sovereign funds. Unlike Google in 2004, when public investors could buy into a company valued near $23 billion, major AI firms now stay private until valuations reach the hundreds of billions or more, leaving ordinary savers with indirect and expensive bets such as Nvidia or Microsoft.
Young workers are being hit first
Entry-level tasks are among the easiest to automate: research, first drafts, simple coding, proofreading, and administrative knowledge work. Research cited from Stanford economist Erik Brynjolfsson and co-authors found a roughly 13% relative employment decline in 2025 among 22- to 25-year-olds in the occupations most exposed to AI, while older workers in those same fields were more resilient.
Middle-class bargaining power is weakening
Even without mass layoffs, the mere possibility of replacing part of white-collar work with AI can slow raises and reduce labor leverage. Roles that retain value are increasingly those tied to legal responsibility, trusted client relationships, and physical presence, while routine execution work becomes cheaper and easier to substitute.
Data created enormous unpaid value
Billions of people helped train AI systems through public text, images, and ongoing software telemetry without receiving compensation. Licensing money largely flows to big platforms such as Reddit, X, image libraries, and major publishers, while broad anonymous contributors receive nothing; even high-profile legal settlements, including Anthropic’s $1.5 billion literary case in 2025, mainly protect large identifiable rights holders.
Profits are shifting away from the models
AI models are being commoditized quickly as training and inference costs fall and open competitors improve fast, especially from China. The most durable rents are moving toward scarce assets around the models: chips, GPUs, energy, data centers, electrical infrastructure, and platforms that control user access and collect telemetry. That helps explain why Google, Amazon, and Apple remain central to the race despite intense attention on standalone model developers.
States face a shrinking tax base
In labor-heavy tax systems, especially in Europe, a shift from wages to capital threatens public finances. If AI-driven profits are booked in the United States, China, or the Middle East, domestic payroll tax receipts weaken while income flows abroad, setting up political fights over digital services taxes, automation levies, and the taxation of computing power.
Households may feel poorer despite gaining services
AI can make intelligence-rich services abundant while leaving scarce physical goods untouched. Housing near jobs, healthcare, schooling, land, and caregiver time remain constrained, so money saved on software and digital assistance may simply be redirected into categories already experiencing inflation, echoing a long U.S. pattern in which televisions and software got cheaper while housing, healthcare, and education surged.
Europe faces a sovereignty risk
Heavy reliance on foreign AI providers could turn advanced models into a dependency similar to imported energy. If European firms buy productivity from American or Chinese platforms through subscriptions and cloud contracts, part of the continent’s value creation is continuously transferred abroad, raising concerns similar to those exposed during the 2022 energy crisis.
The practical route for individuals is scarcity plus results-based pay
The most accessible defense is to stop selling time and start selling outcomes. Fixed-fee contracts, revenue shares, professional signatures, niche knowledge, and AI-augmented physical trades offer better protection than hourly billing. By contrast, reselling generic access to intelligence through prompting courses, thin software wrappers, or mass content production is vulnerable because it sits in the very layer of the stack becoming abundant and cheap.
CONCLUSION
The central divide in the AI economy is not between users and non-users, but between those who consume cheaper intelligence and those who own the scarce assets around it. Unless compensation, taxation, and ownership structures change, AI is likely to keep widening that gap even as it makes daily life more convenient.
Ask a question