Candice Horbacz
As US investment in artificial intelligence (AI) accelerates at an unprecedented pace, a parallel surge has emerged in proposals to uniquely tax this transformative technology. These discussions echo a historical debate from the 1990s concerning the burgeoning World Wide Web and its economic implications. Current AI tax proposals vary widely, with some broadly targeting large corporations and the wealthy, while others are narrowly focused on specific AI activities, such as the "tokens" flowing in and out of AI models or the "compute" resources utilized for their training. This nuanced landscape of proposed AI taxation shares striking conceptual DNA with a controversial 1990s concept: the "bit tax."
The Genesis of the Bit Tax: A 1990s Vision for Digital Revenue
The idea of a bit tax emerged from concerns over the nascent World Wide Web and its potential impacts on global economies and labor markets. In an era before widespread broadband and ubiquitous internet access, policymakers grappled with how to understand, regulate, and potentially monetize the burgeoning digital realm. The original bit tax proposal was put forth by Canadian economist Arthur Cordell, who first presented the concept at a 1995 conference. His remarks at the time, which focused on the economic shifts anticipated from digital transformation, bear a striking resemblance to contemporary conversations surrounding AI. Cordell envisioned a tax applied to the uploads and downloads of information over the internet, a mechanism he believed could address future societal needs and fund the burgeoning global communications infrastructure.
Cordell’s initial suggestion was a remarkably granular tax of 0.000001 cents per bit. To put this into perspective, even a seemingly minuscule rate could accumulate rapidly given the exponential growth of digital data. For instance, a single kilobyte (KB) would incur a tax of $0.000008, a megabyte (MB) $0.008, a gigabyte (GB) $8, and a terabyte (TB) a staggering $8,000. These figures, derived from the standard conversion of bits to bytes (where 1 byte equals 8 bits), illustrate the potential scale of such a levy, even at its proposed low rate.
The rationale behind Cordell’s proposal was multifaceted. He anticipated significant labor displacement due to automation and digital technologies, similar to today’s fears surrounding AI. He argued that such a tax could provide a revenue stream to mitigate these disruptions, perhaps through universal basic income or retraining programs. Furthermore, he saw it as a means to fund essential public services and infrastructure in an increasingly digital world. In a subsequent 1997 speech, Cordell acknowledged the rapid pace of technological advancement, noting that "the bit tax rate will have to be adjusted for changing times." This foresight underscores the inherent challenge of taxing rapidly evolving technologies with fixed, granular rates.
The concept gained international traction in the late 1990s. A 1999 United Nations report, for example, explored a $0.01 per megabyte tax as a potential mechanism to "fund the global communications revolution." The European Commission also delved into the idea, signaling a global interest among policymakers in how to harness the economic potential of the internet while addressing its perceived challenges. Observers at the time often likened the idea to an "email tax," further highlighting its broad applicability to digital communication.
Why the Bit Tax Failed: Complexity, Non-Neutrality, and Anti-Growth Concerns
Despite this initial global interest, the bit tax ultimately never came close to enactment. In the United States, the proposal drew swift and bipartisan opposition, laying bare the fundamental challenges of taxing an emergent and rapidly evolving technology.
A pivotal moment came in 1997 when President Bill Clinton articulated a clear vision, stating his desire to keep the internet "free of new discriminatory taxes." This stance reflected a broader consensus that the internet, as a nascent engine of innovation and economic growth, should not be encumbered by unique levies that could stifle its development. This sentiment solidified into legislative action with the passage of the Internet Tax Freedom Act (ITFA). An early, bipartisan version of ITFA explicitly prohibited states from enacting "bit taxes," signaling a strong legislative aversion to such targeted digital levies.
The bipartisan Advisory Commission on Electronic Commerce, established by the final version of ITFA, further underscored this opposition. In its 2000 report to Congress, the commission concluded that the bit tax was "met with little support by government officials," effectively delivering a decisive blow to the proposal’s viability.
Three primary factors contributed to the demise of the bit tax:
-
Administrative Complexity and Impracticality: Implementing a bit tax would have presented immense logistical challenges. The sheer volume and velocity of data transmission in the digital age made the measurement, collection, and enforcement of such a granular tax extraordinarily complex. Who would be responsible for tracking every bit uploaded or downloaded? Internet service providers? Website hosts? Individuals? The administrative burden on both taxpayers and government agencies would have been staggering, making the proposal impractical to administer effectively. Furthermore, concerns over data privacy and the potential for surveillance through the tracking of digital communications also emerged as significant obstacles.
-
Lack of Neutrality and Discriminatory Application: Critics argued that the bit tax was inherently non-neutral. It would have singled out digital transactions for taxation in a way that physical goods and services were not, creating an uneven playing field. The principle of tax neutrality dictates that taxes should neither favor nor penalize particular economic activities, industries, or forms of commerce. By imposing a unique tax on digital bits, the proposal was seen as discriminatory against the burgeoning internet economy, potentially distorting market behavior and hindering its natural development. It would have acted as a "toll on internet usage," impacting everything from email to early web browsing.
-
Anti-Growth and Innovation-Stifling Potential: Perhaps the most significant concern was the potential for the bit tax to stifle innovation and economic growth. The internet in the 1990s was in its infancy, and many feared that imposing a tax on its fundamental units of transmission would deter investment, discourage usage, and impede the development of new technologies and services. The argument was that such a tax would make data-intensive activities more expensive, thereby limiting their adoption and growth. Had the bit tax been enacted and maintained as originally proposed, it would have either soaked consumers and businesses in extraordinarily high taxes today or actively prevented the development of new, data-intensive technologies altogether, fundamentally altering the trajectory of the digital revolution.
Lessons for the AI Era: Rapid Technological Evolution
Although the bit tax debate occurred nearly three decades ago, its lessons are remarkably pertinent for policymakers considering AI taxes today. The most salient lesson is the rapid and unpredictable pace of technological change. A bespoke tax mechanism that might seem sensible in one era can quickly become obsolete, disproportionately burdensome, or even nonsensical in the next.
Consider the example of data usage. In the 1990s, Cordell’s proposed rate might have seemed negligible. However, today, the median US household uses an astonishing 532 GB of data per month. If Cordell’s original bit tax of $8 per gigabyte were applied to current internet usage, the average American household would face an additional monthly tax bill of over $4,256 – an extraordinarily high and prohibitive levy. This scenario vividly illustrates how a fixed tax rate on a rapidly expanding digital metric can quickly become unfeasible.
Few in the 1990s could have envisioned seamless global video calling, on-demand streaming of thousands of movies, or the complex, data-intensive applications commonplace today. Similarly, a "compute" tax on processing power or a "token" tax on AI model outputs, measured today, could become equally nonsensical in a mere two, five, or ten years. As AI models become more efficient, larger, or fundamentally change their operational metrics, a fixed tax based on current parameters could either become negligible, rendering it ineffective for revenue generation, or become cripplingly expensive, stifling the very innovation it aims to regulate or tax.
For instance, applying a hypothetical bit tax (using Cordell’s original rate) to common internet uses today demonstrates this absurdity. A single high-definition movie stream (typically 3-7 GB) could incur a tax of $24 to $56. A 30-minute video call (around 0.5-1 GB) would cost $4 to $8 in tax. Even simply browsing the web for an hour (approximately 0.1 GB) would add $0.80 to the bill. These figures, while hypothetical, underscore how rapidly the burden would escalate with modern data consumption, an outcome Cordell certainly did not intend.
Lessons for the AI Era: The Imperative of Neutrality
Another timeless principle that emerged from the bit tax debate is the importance of tax neutrality. In the 1990s, a bipartisan coalition of executive and legislative branch officials in the US staunchly defended the principle that internet products and services should neither face discriminatory taxes nor gain an unfair advantage over traditional "brick-and-mortar" goods and services.
This principle guided the resolution of the online sales tax debate, which saw states and the federal government wrestle with the "online advantage" for decades. The issue was ultimately settled by the Supreme Court’s Wayfair decision in 2018, which affirmed states’ rights to require remote sellers to collect sales tax, thereby applying the principle of neutrality to ensure equitable competition between online and physical retailers.
Today, some AI tax proposals are framed as broad tax increases on wealth or capital, which, while debatable on other grounds, might be considered more neutral if they apply broadly across the economy. However, many proposed AI taxes are narrowly targeted at specific AI companies or activities, raising concerns about non-neutrality. Taxing "compute" used for AI, for example, but not compute used for other advanced data processing or scientific research, could create an uneven playing field. Such targeted taxes risk penalizing innovation in AI specifically, potentially hindering its development relative to other technologies or industries. Policymakers today should adhere to the same principles of neutrality that led to the rejection of the bit tax in the 1990s, ensuring that AI is neither unfairly burdened nor unduly favored by the tax code.
Lessons for the AI Era: Guarding Against Stifled Growth
The potential adverse impact on economic growth and innovation is a third crucial lesson from the bit tax episode. Had the US enacted a bit tax, it is conceivable that Americans today might not enjoy the widespread benefits of remote work, accessible telehealth services, or the broader social connectedness fostered by high-speed internet. Even at a much lower rate than Cordell’s original proposal, a bit tax would have inevitably slowed activities that drive job creation and economic expansion by increasing the cost of digital interaction.
Contrary to initial fears, the labor displacement Cordell warned of in the 1990s ultimately did not materialize on a catastrophic scale. Instead, the rise of the internet ushered in a new era of economic growth, creating entirely new industries, jobs, and services. This digital revolution even brought with it a significant financial and tax revenue boom that, albeit temporarily, improved the US economic and budgetary outlook in the late 1990s and early 2000s, demonstrating the immense economic dividends that can arise from fostering technological innovation rather than immediately taxing it.
Applying this lesson to AI, policymakers must weigh the potential for targeted taxes to stifle an emerging technology with vast potential. While AI undeniably presents significant disruptive challenges, including concerns about job displacement, it also holds the promise of unprecedented productivity gains, new industries, and solutions to complex global problems. Imposing unique, burdensome taxes on AI development or deployment could impede its beneficial applications, deter investment, and slow the pace of innovation, potentially costing the economy more in lost growth than it gains in revenue.
The Nuances of the Modern Landscape
It is crucial to acknowledge that the 2020s are not merely a mirror image of the 1990s. The economic and technological landscapes differ significantly. US economic growth is generally slower today than it was three decades ago, and the national budget is considerably more vulnerable to major threats to its tax base, with a national debt that has swelled significantly. Furthermore, AI could prove to be a more powerful and disruptive technology than the World Wide Web, potentially introducing greater societal and economic shifts. These differences mean that the challenges and considerations for AI taxation are distinct in some respects.
However, the core tax principles highlighted by the bit tax debate remain timeless: simplicity, neutrality, transparency, and stability. Policymakers evaluating complex or non-neutral AI tax proposals in the 2020s should consider the 1990s bit tax a cautionary tale. While the desire to generate revenue, address societal concerns, or regulate powerful new technologies is understandable, history suggests that targeted, granular taxes on the fundamental elements of nascent technologies can have unintended and detrimental consequences, potentially stifling the very innovation that promises future prosperity. Learning from the failure of the bit tax can serve as a vital guide, urging a thoughtful, principles-based approach to ensure that AI’s transformative potential is harnessed, not hampered, by ill-conceived taxation.
