Amid growing anxiety that rapid advancements in artificial intelligence could trigger substantial and swift disruptions across domestic labor markets, Senator Mark Kelly (D-AZ) has introduced a legislative package aimed at mitigating the fallout. The proposed measure, titled the Make AI Work for Americans Act, seeks to inject crucial federal funding into various labor insurance and worker retraining programs. While the precise long-term trajectory of artificial intelligence and its ultimate impact on the labor market remain subjects of ongoing debate among economists and policymakers, some prominent industry observers project that millions of workers may be forced into entirely new career paths as automation reshapes traditional job descriptions.
To finance these newly proposed federal expenditures, Senator Kelly’s bill introduces a trio of distinct tax mechanisms: a digital advertising tax, a bit tax levied on data usage, and an excess profits tax targeting highly lucrative firms. Although the legislative effort attempts to address the pressing necessity of paying for new federal spending—particularly at a time when the national debt burden and associated interest payments sit near record highs and are projected to climb further—analysts warn that the structure of the proposal could inadvertently worsen federal deficits. This risk arises because the bill spends gross revenue projections without properly accounting for anticipated declines in other vital federal revenue streams, such as individual income and payroll taxes.
Revenue Estimates and Implementation Challenges
Should the legislation pass, all three proposed taxes would officially take effect on January 1, 2027. Economic modeling indicates that, assuming no immediate behavioral changes or shifts in broader economic growth resulting from the new levies, the digital advertising tax would generate approximately $15.8 billion in its inaugural year. Meanwhile, the excess profits tax is estimated to bring in $6.5 billion, while the bit tax would yield a comparatively modest $23.2 million during the same timeframe.
Projecting tax revenues over a ten-year horizon presents significant challenges, largely because the growth rates of the affected economic sectors remain highly uncertain. Furthermore, because these preliminary estimates do not incorporate taxpayer behavioral responses—such as aggressive tax avoidance strategies—nor do they account for broader macroeconomic feedback effects, the figures function effectively as a ceiling rather than a guaranteed outcome. While a combined collection of roughly $380 billion over a decade is theoretically possible under static conditions, actual receipts would likely fall well short once economic adjustments are factored into the equation.
Historical Precedents and Practical Concerns
While Senator Kelly appears to be the first federal lawmaker to package these three specific levies together into a single cohesive bill, none of the underlying concepts are entirely novel. Each tax possesses a distinct historical precedent at the federal, state, or international levels.
Digital advertising taxes, for instance, have gained considerable traction at the state level in recent years. Jurisdictions such as Maryland, Illinois, and Utah have all enacted targeted digital ad taxes, though the policy has faced severe legal hurdles. Most notably, a Maryland state tax court struck down that state’s digital ad tax in a widely watched ruling, prompting an ongoing state appeal.
The bit tax carries an even longer history, tracing its conceptual roots back to policy discussions from the 1990s. Economists note that a 1990s-era bit tax applied to today’s data volumes would amount to an astronomical $80 per gigabyte—an outcome certainly not intended by its original proponents, but a stark reminder of how rapidly technology outpaces rigid tax metrics. Under Senator Kelly’s current proposal, the bit tax would be set at $11 cents per gigabyte. While that rate may appear negligible today—given that a standard text-based interaction with an artificial intelligence platform consumes only a fraction of a megabyte—the exponential expansion of data usage over the next decade could transform the levy into a prohibitive barrier against technological development.
Excess profits taxes also possess a well-worn legislative history, frequently surfacing at the federal level during periods of soaring energy prices under the guise of windfall profits taxes on oil and gas producers. Historical analyses of the 1980 Windfall Profits Tax, for example, demonstrated that the policy significantly reduced domestic energy production while driving up reliance on foreign imports. Furthermore, the 1980 tax brought in a mere fraction of its initial revenue projections, collecting roughly $80 billion compared to an expected $393 billion. Senator Kelly’s proposed excess profits tax similarly suffers from an exceptionally narrow base, potentially falling almost entirely on a single corporate taxpayer and rendering the revenue stream precarious.
Structural Flaws in the Proposed Tax Regime
From the perspective of traditional public finance, sound tax policy should remain neutral, simple, transparent, and stable. In contrast, the taxes outlined in the Make AI Work for Americans Act are narrowly targeted, complex to administer, and dependent on tax bases susceptible to wild fluctuations driven by technological innovation or avoidance behavior.
The digital ad tax operates as a gross receipts tax levied on a standard business input. Similar digital service taxes attempted abroad have drawn strong opposition from the United States government due to their tendency to distort production, pyramid through complex supply chains, encourage inefficient corporate restructuring, and penalize low-margin enterprises. Additionally, because many human creators—such as authors, artists, and journalists—ultimately rely on digital advertising revenue to monetize their work, a tax on digital ads would partially burden the very creative workforce that faces displacement or disruption from emerging artificial intelligence tools.
The structural design of the bit tax similarly threatens to impede market growth. If artificial intelligence adoption follows a trajectory comparable to the explosive expansion of the internet—where data usage soared while unit costs collapsed—a fixed tax per gigabyte would rapidly escalate into a crushing financial burden for firms attempting to integrate modern technologies. While shifting toward an ad valorem approach might alleviate the danger of an escalating nominal burden, it would still distort economic decision-making away from efficient technological inputs.
Finally, the excess profits tax applies a remarkably high rate to an unusually narrow base. By singling out a handful of successful firms based on arbitrary financial conditions, the policy violates principles of tax neutrality and deters future innovation by imposing intertemporally asymmetric treatment on successful research and development investments.
Lawmakers striving to fund new domestic initiatives or address long-term federal deficits would achieve better outcomes by pursuing broad-based reforms, such as broadening income tax bases, reforming the corporate income tax structure, or shifting reliance toward consumption-based taxation.
