When capital can think, who pays?

<p><b>AI is remaking the economy. Incentives that shape its deployment should ensure more people share in what it creates.</b></p>

<p><i>This article was originally published by </i><a href="https://www.newsweek.com/when-capital-can-think-who-pays-opinion-11759860" target="_blank"><i>Newsweek</i></a><i> in April 2026.</i></p> <p>What happens when society embraces a technology faster than it can absorb its consequences? In the United States, AI adoption is growing at a remarkable pace. As of December 2025, 32% of large US firms had integrated AI into production or core administrative workflows, roughly double the rate from a year earlier. Private, non-work use of AI hit 48.7% in late 2025, a 35% growth off already one of the highest individual adoption rates in the world.</p> <p>Yet the public mood is not one of enthusiasm: Americans are far more concerned than excited about AI’s role in daily life. Headlines about sizeable layoffs tied to automation and robotics have become a regular feature of leading firms’ earnings reports, and investors increasingly expect more to come.</p> <p>While AI’s role in job displacement may be overstated in parts of the media, the underlying data suggests disruption may be arriving faster and reaching deeper than anticipated. Our <a href="https://www.cognizant.com/us/en/aem-i/ai-and-the-future-of-work-report" target="_blank">New work, new world 2026</a> research, a re-assessment of 18,000 tasks across nearly 1,000 professions, finds that, based on AI’s capabilities today, 93% of jobs are exposed to some degree of AI-led automation, with exposure scores climbing at 4.5 times the anticipated rate and the share of jobs with minimal exposure collapsing from 31% to 7%.</p> <p>If AI is beginning to disrupt work faster than new jobs have time to emerge, how should public policy support a more equitable transition phase, without inhibiting innovation? One answer: temporarily rebalance the cost of automation so that using AI to augment human workers is at least as attractive as using it to replace them.</p> <p>A temporary shift in how automation is taxed relative to human labor could help preserve the link between value creation and broad-based prosperity during the most consequential transition of our time.</p> <h4>When capital can think</h4> <p>For decades, governments taxed labor more heavily than capital because the underlying assumption was that capital investment would expand jobs, wages and broader prosperity. Economic systems were built for intelligence scarcity: a world where machines needed human judgment to be productive and returns on investment flowed back into employment. The United States is not an outlier in this regard. The entire fiscal architecture of the developed world rests on the assumption that capital creates jobs and workers fund the lion's share of the state.</p> <p>AI upends the equation. Intelligence is now becoming abundant, and the value created from it is accruing more quickly to capital than to labor. Yet the current tax system still favors capital versus labor, treating automation as productive investment and employment as a cost. The logic of the old-world arrangement was simple: we did not tax capital heavily because the value it created flowed to workers. Now, the value increasingly flows to capital owners—yet we still tax the workers.</p> <p>Consider what this looks like in practice. A digital agent performing the same tasks as a human worker pays nothing into the system that is funded by payroll taxes—no Social Security, no unemployment insurance, no Medicare. This results in a growing financial incentive for firms to automate rather than hire.</p> <p>As AI changes where economic value is created and captured, tax policy should evolve to reflect this shift.</p> <p>If capital is taxed more and labor less, replacing people with AI is no longer the cheapest path, and using AI to augment human workers rather than replace them becomes a more attractive option.</p> <p>With the incentive structure less tilted toward labor replacement, firms will have more reason to use AI to create new tasks and raise worker output, thus preserving the connection between value and human work and making broad-based prosperity more likely.</p> <h4>The cost of waiting</h4> <p>What makes this moment structurally different from past technology revolutions is the combination of AI’s extreme capital intensity and its inverted labor disruption. The railroads, built over two decades, hovered around 2.5% of GDP in annual capital expenditure. The telecom and fiber build-out of the late 1990s reached about 1.5%. The internet era&nbsp;<a href="https://www.frbsf.org/wp-content/uploads/er19-34bk.pdf" target="_blank">peaked around 1.2%</a>. AI infrastructure spending has already hit 1% of GDP and we are still in the early stages, on a&nbsp;<a href="https://www.apolloacademy.com/how-much-is-646-billion/" target="_blank">trajectory toward 2% or more</a>. This is shaping up to be the most capital-intensive technology transition in modern history, and the gains from that investment are flowing disproportionately to those few who own the infrastructure behind it.</p> <p>At the same time, AI is causing an inverse labor disruption compared to every prior technology wave. Previous revolutions displaced manual labor first. AI inverts the pattern, reaching white-collar, knowledge-work first, and falling hardest on the part of the labor market that drives the highest wages, consumer demand and tax revenue.</p> <p>Left alone, this combination can become self-reinforcing, with more capital funding more labor displacement, which in turn generates more returns to capital, which funds more automation. Over time, this can concentrate gains and strain the labor tax base, eroding the revenue streams that underpin the broader economy—funding health care, retirement benefits and basic economic security.</p> <h4>A catalyst for a more inventive transition</h4> <p>The objections to a shift towards higher taxation of capital are not to be dismissed. AI may, over time, create more jobs than it destroys, as past technological revolutions did. Taxing capital too aggressively could also slow down innovation or cause investment to migrate to other jurisdictions. And at 4% unemployment, one could argue the labor market is healthy enough.</p> <p>But that headline number masks structural shifts: from 1979 to 2025,&nbsp;<a href="https://fred.stlouisfed.org/data/OPHNFB" target="_blank">labor productivity grew</a>&nbsp;about 2.5 times as fast as&nbsp;<a href="https://fred.stlouisfed.org/data/COMPRNFB" target="_blank">real hourly compensation</a>. The displacement is now hitting the highest value segment of the workforce.</p> <p>More importantly, these objections assume a permanent ideological shift. What is proposed here is something different: a temporary bridge policy to enable a more equitable transition during the time when the jobs of the past are disappearing and the jobs of the future have not yet arrived. The purpose of a temporary tax rebalancing is not to suppress AI but to encourage its use as a catalyst for a better, more inventive transition.</p> <p>Constraint has often been the mother of invention. If replacing workers is no longer the cheapest path, firms will have stronger reason to discover how AI can amplify human capability, raising output, creating new jobs and enabling people and machines to generate value that neither could produce alone. History suggests that it’s possible: 60% of today’s occupations did not exist in 1940. New jobs are born from new tasks, and new tasks emerge when the incentive structure rewards creation over elimination.</p> <p>Work is still how people earn security, build dignity and participate in growth. As AI is remaking the economy, the incentives shaping its deployment should ensure that more people share in what it creates.</p>