Faster growth may no longer translate into equal tax revenue growth

Key takeaway

Artificial intelligence can accelerate economic growth without producing an equal increase in government revenue. AI may raise productivity and profits while shifting value from wages to capital income, from domestic firms to global platforms and from currently taxed earnings to profits taxed only when distributed.

BTU researchers assess that Georgia’s main risk is not an immediate fiscal crisis but a structural change in the tax base. Public debt is low and growth remains strong, yet a large share of 2025 state tax revenue depended on personal income and consumption. If the benefits of AI accumulate mainly in profits while employment and wage growth weaken, tax receipts may lag behind the wider economy.

Why the global debate matters for Georgia

The Economist highlights a fiscal paradox: AI may expand productivity, corporate earnings and GDP while reducing the share of income that governments tax most effectively. In advanced economies, the immediate concern is the heavy reliance on labour taxes and social-security contributions.

Georgia does not replicate the high-tax OECD model. It has no conventional high payroll contributions and receives a large share of revenue from VAT. That provides some protection, but not immunity. Personal income tax is one of the two largest revenue sources, while VAT ultimately depends on household consumption. Weaker labour income can therefore affect both channels.

What the 2025 budget structure shows

Georgia’s state budget collected approximately GEL 21.96 billion in tax revenue in 2025. Personal income tax generated GEL 7.52 billion, VAT GEL 8.13 billion, profit tax GEL 2.88 billion and excise duties GEL 2.73 billion.

BTU researchers calculate that personal income tax represented 34.2% of total state tax revenue, VAT 37.0% and profit tax only 13.1%. Personal income tax, VAT and excise duties together accounted for 83.7%. Not all of this is mechanically tied to wages, but the structure demonstrates how strongly the budget depends on employment, labour income and everyday consumption.

Georgia’s distinctive tax design

Georgia’s standard personal income tax rate is 20%, VAT is 18% and the profit tax rate for most companies is 15%. Corporate profit is generally taxed upon distribution. This design encourages investment because retained and reinvested earnings are not immediately taxed.

In an AI economy, that advantage also creates a timing risk. Automation may raise corporate earnings, but if profits are retained for long periods, economic value is created today while the corresponding profit-tax revenue reaches the budget later. Wage tax is collected monthly; tax on reinvested profit may arrive years afterward. A shift from labour to capital can therefore create a timing problem as well as a revenue-composition problem.

BTU stress test

Using actual 2025 receipts, BTU researchers constructed a simple static stress test. If personal income tax revenue were 5% below the baseline because of AI adoption, changing work arrangements and weaker wage growth, the direct gap would be about GEL 376 million. At a 10% shortfall, the gap would approach GEL 752 million.

Those amounts equal approximately 1.7% and 3.4% of total 2025 state tax revenue. This is not a forecast and does not claim that AI will necessarily cause such losses. It is a sensitivity calculation showing how quickly weakness in one major tax source could become material. Faster GDP growth, profit tax and VAT may offset part of the gap, but that compensation is not automatic.

Where value is created and where it is taxed

Foreign AI platforms create a particular challenge for Georgia. A Georgian company may save local labour costs while paying more for a service supplied from abroad. Productivity rises in Georgia, but much of the technology profit remains with a provider registered elsewhere. The domestic wage-tax base can weaken without an equivalent corporate-profit base appearing locally.

VAT captures part of imported digital services, but it cannot substitute for profit taxation. The policy question is therefore broader than rates. Georgia needs accurate recording of digital transactions, effective source and permanent-establishment rules, transfer-pricing enforcement and international tax cooperation. A small open economy gains little from a heavy unilateral levy if the underlying income can easily move across borders.

Why a direct AI tax is not the best first step

A tax on computing power, tokens or software use may sound simple, but it would carry large costs for Georgia. The country is still at an early stage of broad AI adoption, and a new levy would make the digital transformation of local firms more expensive. The domestic AI spending base is also unlikely to be large enough to replace meaningful losses from labour and consumption taxes.

BTU researchers assess that Georgia should tax the economic results created by AI rather than the use of the technology itself. The priorities are neutral treatment of income across different forms, review of unnecessary tax relief, comprehensive reporting of digital services and stronger protection against artificial profit shifting. This approach preserves innovation incentives while defending a broader revenue base.

AI also changes the spending side

The fiscal effect is not limited to revenue. Workers affected by automation may require retraining, temporary income support and new employment services. At the same time, faster wage growth in highly productive private activities may increase the cost of retaining skilled staff in government. Revenue growth could slow while education and social-policy needs rise.

AI can also raise public-sector productivity. Tax-risk detection, customs control, targeting of social programmes, health planning and administrative services can be delivered more quickly and at lower cost. Georgia’s fiscal strategy should therefore not focus only on finding new revenue; using existing public money more effectively is equally important.

What Georgia should do now

The first step is a data system linking AI use with employment, wages, corporate profits and taxes by sector. GDP growth alone cannot show whether taxable labour income is keeping pace. The second is to include changes in labour’s income share in medium-term fiscal scenarios and regularly test the sensitivity of personal income tax.

The third is tax neutrality. The state should neither tax human labour disproportionately nor punish technology investment. Policy should focus on a broad base, effective administration, fair allocation of international profits and support that helps workers move into more productive activities.

Conclusion

AI adoption can expand Georgia’s economy while making its budget more difficult to finance. The 2025 revenue structure shows continued reliance on personal income and consumption, while profit tax contributes a much smaller share. If value moves from wages to profits, tax receipts may fail to keep pace with economic growth.

BTU researchers conclude that the best response is not to tax AI itself but to prepare the tax system before the shift accelerates. Georgia needs better measurement, fiscal scenarios, stronger reporting of digital income, protection against profit shifting and investment in human capital. The objective is to ensure that AI-driven productivity expands not only private profit but also the country’s fiscal capacity.

Data and Main Sources

The Economist — How the AI Boom Could Worsen the Rich World’s Fiscal Crunch

Parliamentary Budget Office of Georgia — 2025 State Budget Execution

Tax Code of Georgia

International Monetary Fund — Georgia 2026 Article IV Consultation

Ministry of Finance of Georgia — Medium Term Revenue Strategy

Disclaimer

This material is analytical and educational. It does not constitute financial, investment, tax or legal advice.

Prepared by the academic team of Business and Technology University and the BTUAI Research Team, Tbilisi, Georgia.