Tom Blomfield, founder of the digital bank Monzo, has made a bold prediction: traditional income tax could become obsolete within five years as advancements in artificial intelligence (AI) reshape the economic landscape. He argues that instead of taxing human labor, governments should consider taxing computational infrastructure, such as data centers.
Blomfield stated, “I don’t think we’ll tax human labour, we’ll tax compute, [meaning systems like] data centres, and then we will use the proceeds to pay for government.” This shift comes as AI systems are increasingly outperforming humans in specific tasks, with some tools even surpassing university professor-level performance.
Currently, the UK government relies on income tax and National Insurance for a staggering 42% of its revenue, while capital-based taxes contribute a mere 4%. As AI continues to evolve, it is predicted that tax accounting could soon require virtually no human involvement, further complicating the future of income tax.
Blomfield emphasized that while AI tools are not yet generalizable, they are expected to become so by the end of 2026. He remarked, “They’re not yet generalisable, so they’re very narrow geniuses, but by the end of 2026 they will be generalisable.” This evolution could lead to significant changes in how governments approach taxation.
The implications of such a shift are profound. With a 35% drop in entry-level job advertisements since the launch of ChatGPT, the Labour Party government may soon need to levy taxes on computing power as automation increasingly replaces human workers. Morgan Stanley has warned that the UK could face an AI-driven employment crisis more acutely than other nations, raising urgent questions about the sustainability of the current tax system.
As the landscape of work and taxation evolves, observers are left to ponder what the future holds. The biggest ever change to HMRC income tax is reportedly coming “within five years.” Details remain unconfirmed, but the urgency of adapting to these changes cannot be overstated.