Figure 6.3 Since the launch of GPT-1 in 2018, the number of firms engaged in AI-related lobbying in the United States has rapidly increased Number of firms, by topic 2,000 1,500 1,000 500 19 19 98 9 20 9 00 20 0 20 1 02 20 0 20 3 04 20 0 20 5 06 20 0 20 7 08 20 0 20 9 10 20 1 20 1 12 20 1 20 3 14 20 1 20 5 16 20 1 20 7 18 20 1 20 9 20 20 21 20 22 20 2 20 3 24 20 25 0 Immigration Science and technology Source: WDR 2026 team, based on data of Top Issues (dashboard), OpenSecrets, https://www.opensecrets.org/federal​ -lobbying/top-issues. Note: The year 2018 corresponds to the launch of the first OpenAI large language model, Generative Pre-trained Transformer 1 (GPT-1). The plotted lines show the total number of firms engaged in lobbying by issue. The AI-related category science and technology is the combined total of science and technology, computers and information technology, and manufacturing. Lobbying on immigration is shown separately. If a firm lobbies on more than one issue, it is counted multiple times in the data. AI = artificial intelligence. In developing countries, the power imbalance between firms and governments is often exacerbated by weak state capacity. This implies that even for countries that navigate the geopolitics well and can secure access to various layers of the AI stack, the opportunities for AI to transform their economies can still be wasted because of the ways in which domestic governments engage with AI firms. Three outcomes are highlighted next. Each lets the opportunity slip away in different ways: straining government finances, foreclosing the investment that is needed, or channeling the gains to those who already hold power. Competition to attract AI investment strains government finances and maneuver room When firms are able to secure significant concessions from governments, competition among 208 countries to attract AI-related investments may lead some countries to commit significant fiscal resources. Examples of fiscal commitments abound. In Uzbekistan, the government has introduced policies to attract investment in AI and data centers through tax exemptions and discounted electricity rates.32 India has announced a 20-year tax holiday until 2047 for foreign providers of cloud services and data center operators.33 While these arrangements may incentivize AI investments, if entered into excessively, they may place a disproportionate share of the financial and technological risks on the public while allowing firms to capture a significant share of the economic returns. Fiscal balances in many developing countries are already deteriorating while pressure from rising debt is growing.34 Such policies risk creating a race to the bottom in which countries undercut one another in offering incentives to attract investments, further straining fiscal space. World Development Report 2026