WASHINGTON — Prompted by growing concerns that the rapid adoption of artificial intelligence could trigger substantial and disruptive shifts in the modern labor market, Senator Mark Kelly (D-AZ) has introduced legislation aimed at ensuring major technology companies contribute directly to worker support programs. The proposed legislation, titled the "Make AI Work for Americans Act," seeks to channel federal funding into an array of labor insurance frameworks and specialized worker training initiatives designed to cushion the blow of automation.
While the exact long-term trajectory of AI’s impact on the labor market continues to be a subject of intense debate among economists and policymakers, various labor market observers and forecasting groups project that millions of workers may ultimately be forced to transition into entirely new professional roles. Senator Kelly’s legislative package attempts to address this displacement head-on by establishing new revenue streams dedicated specifically to workforce resilience. To fund these ambitious spending programs, the proposal introduces three distinct tax mechanisms: a digital advertising tax, a bit tax on data utilization, and an excess profits tax targeting high-earning entities within the technology sector.
Senator Kelly’s willingness to identify specific revenue sources to pay for new spending has drawn praise from fiscal analysts. Amid an era where the federal government’s debt burden and associated interest payments sit near record highs—with projections pointing toward continued fiscal expansion—policymakers are increasingly urged to avoid proposing unfinanced federal spending. However, a closer examination of the senator’s proposal suggests that the revenue structure may inadvertently exacerbate federal deficits rather than reduce them. The legislation accounts for gross revenues generated by the new levies, yet it fails to factor in anticipated declines in existing federal tax
receipts, such as income and payroll taxes, that would likely result from the imposition of these industry-specific burdens.
Under the legislative timeline, all three proposed taxes would take effect on January 1, 2027. Static fiscal estimates—which do not account for potential behavioral shifts, avoidance strategies, or broader macroeconomic adaptations by affected firms—indicate that the digital advertising tax would raise approximately $15.8 billion in its first year alone. Meanwhile, the excess profits tax is projected to generate about $6.5 billion, and the bit tax would yield roughly $23.2 million during the same initial twelve-month period.
Projecting revenue over a ten-year horizon presents significant challenges due to the inherent volatility and uncertainty surrounding the growth rates of the underlying tax bases. Furthermore, because these estimates omit behavioral responses, corporate tax avoidance, and macroeconomic feedback loops, they function more as theoretical ceilings than reliable fiscal yields. Over a full decade, the combined package could theoretically generate up to $380 billion, but this figure overstates actual expected revenue because it lacks any adjustment for how taxpayers would inevitably alter their operations to minimize liability.
Fiscal policy experts argue that all three components of the tax proposal are fundamentally unsound. The digital advertising tax functions as a gross receipts levy that can be assessed multiple times at various stages along the advertising supply chain. The bit tax risks becoming an increasingly prohibitive share of the operational cost of utilizing AI systems, thereby serving as an unexpected brake on technological development. Finally, the excess profits tax is structured in a manner that means its financial burden would fall disproportionately, or perhaps almost entirely, on a single taxpayer. Economists suggest that if lawmakers genuinely wish to finance new federal programs, broad-based consumption taxes would achieve those revenue goals at a substantially lower economic cost.
Historical Precedents
While Senator Kelly may be breaking new ground by packaging these specific three taxes into a single legislative vehicle, none of the underlying concepts are entirely novel. Each proposal carries a distinct historical precedent at either the federal, state, or international level.
Digital advertising taxes have found their greatest popularity among state-level policymakers in recent years. States such as Maryland, Illinois, and Utah have enacted targeted taxes on digital ad services. However, these state-level experiments have faced severe legal hurdles. Most notably, Maryland’s digital advertising tax was struck down by the state’s Tax Court in a closely watched legal challenge, prompting an immediate appeal by the state government.
The bit tax boasts an even longer pedigree, tracing its intellectual origins back to discussions from the 1990s. When adjusted for modern standards, a 1990s-era bit tax would equate to roughly $80 per gigabyte today. While such a high rate was likely never the original intention of its early proponents, it serves as a clear historical warning that tax structures conceived in one technological era can become entirely impractical as digital capacities expand.
Senator Kelly’s proposed bit tax amounts to $0.11 per gigabyte. While this rate may appear negligible by today’s standards—given that a typical conversational interaction with an artificial intelligence platform consumes only a few megabytes per hour—the rapid scaling of technology means the burden could escalate dramatically over the coming decades. As AI platforms grow more sophisticated and data-intensive, a fixed tax per gigabyte could transform into a prohibitive barrier for everyday users and enterprises alike.
Excess profits taxes possess an even more extensive and controversial history in American public policy. Federal lawmakers have frequently turned to proposals for excess or windfall profits taxes on oil and gas producers during periods of rising energy prices. Historical analyses of the 1980 Windfall Profits Tax, for instance, concluded that the measure ultimately reduced domestic energy production and increased the nation’s reliance on foreign imports while collecting only a fraction of the revenue originally projected. Senator Kelly’s proposed excess profits tax similarly features an exceptionally narrow base, concentrating its impact on a minuscule number of firms, which makes its revenue yield highly precarious.
The Ad, Bit, and Excess Profits Taxes Are Categorically Unsound
From the perspective of public finance, sound tax policy must remain neutral, simple, transparent, and stable. The taxes proposed in the Make AI Work for Americans Act violate several of these core tenets by being narrowly targeted, complex to administer, and reliant on tax bases that are highly susceptible to technological disruption and avoidance maneuvers.
The digital ad tax operates as a gross-based levy on a fundamental business input. The United States government has historically opposed similar digital service taxes implemented by foreign jurisdictions abroad. Gross receipts taxes are widely recognized by economists as distortionary instruments that pyramid through supply chains, penalize low-margin enterprises, and encourage inefficient corporate restructuring, such as vertical integration.
Furthermore, the tax intersects awkwardly with ongoing debates regarding intellectual property and content creation. Numerous academics and lawmakers have expressed deep concerns that modern AI systems are trained on the uncompensated labor of human creators, including authors, journalists, and artists. A parallel economic concern is that AI-generated search summaries and chat interfaces divert web traffic away from original publisher sites, diminishing monetization opportunities and threatening the production of human-created content. Because many digital creators ultimately rely on advertising revenue for their livelihoods, a digital ad tax would partially fall on the very human workers and creators whose economic security the legislation seeks to protect.
The structural design of the proposed bit tax mirrors the problematic models of the 1990s. If internet and AI usage scales in a manner comparable to the early decades of the consumer internet—where traffic doubled annually while unit costs plummeted—a fixed tax per unit of data would quickly morph from a minor levy into an immense financial burden. While converting the bit tax to an ad valorem basis, similar to token tax proposals introduced by other members of Congress, could alleviate the risk of an escalating absolute burden, it would still distort economic choices by discouraging the integration of AI models in favor of less efficient inputs. Moreover, a bit tax would introduce formidable administrative complexities, requiring regulatory policing of technical avoidance strategies such as data routing and service offshoring.
The excess profits tax compounds these structural flaws by applying a high marginal rate to an exceptionally narrow tax base. Economic analyses indicate that the tax would likely be absorbed by only a handful of firms, with virtually the entire liability falling on a single corporation in its initial year. Singling out specific companies for punitive taxation based on arbitrary financial thresholds violates the principle of tax neutrality. It also creates powerful incentives for corporate restructuring and financial engineering designed to bypass the levy. To the extent that the tax is successfully collected, it penalizes the payoffs to successful innovation at an elevated rate without providing equivalent relief for the exploratory costs incurred earlier in the development cycle, thereby dampening long-term research and development incentives.
Big Picture
Although Senator Kelly’s legislative initiative stands out for attempting to identify offsetting revenue sources for new federal spending, the specific tax mechanisms chosen in the Make AI Work for Americans Act remain problematic. The digital advertising tax is economically inefficient and administratively cumbersome, risking the taxation of revenues bearing little connection to artificial intelligence itself. The bit tax risks obsolescence as digital consumption scales upward, and the excess profits tax relies upon a foundation so narrow that the strategic decisions of a single firm could render the revenue stream moot.
Fiscal analysts maintain that lawmakers seeking to finance new domestic priorities or address broader national debt trajectories would achieve more sustainable results by broadening income tax bases, reforming corporate income taxation, or shifting toward consumption-based levies. Comprehensive options for modernizing the federal tax code remain widely available in professional policy literature for lawmakers aiming to balance fiscal responsibility with economic growth.
No comments yet. Be the first to join the conversation!