Can AI trigger super growth in the global economy?

Artificial Intelligence (AI) is becoming one of the most important variables in global economic growth. From the predictions of Silicon Valley optimists to the calculations of mainstream economic institutions, whether AI can push the global economic growth rate from the current 2-3% to an "explosive" level of 20-30% has become a core topic of intense debate across all sectors. This article provides a systematic analysis by combining historical economic growth trajectories, economic theory models, the latest investment and energy data from 2026, and real-world bottlenecks.

A Historical Perspective: Shifts in Growth Paradigms from Stagnation to Acceleration

Before 1700, the global economy grew at an average annual rate of only about 0.1%, essentially stagnating. After the Industrial Revolution, technological breakthroughs such as the steam engine raised the growth rate to 0.5% between 1700 and 1820, and further to 1.9% by the late 19th century. In the 20th century, global output grew at an average annual rate of 2.8%. This long-term trend shows that technological innovation, by improving productivity and capital accumulation, has achieved a stepwise leap in growth rates.

AI is seen as a General Purpose Technology similar to or even surpassing the Industrial Revolution. Unlike previous technologies, AI has the potential for self-iteration and can automate the vast majority of cognitive and physical tasks, thereby achieving an exponential acceleration of "labor accumulation." This stands in stark contrast to historical "population accumulation": traditional growth relied on generational replacement, while AI "workers" can be rapidly replicated through investment.

Theoretical Mechanisms of Explosive AI Growth

Mainstream economic growth models often predict explosive growth when assuming that AI can effectively replace human labor. Both semi-endogenous growth models and exogenous growth models show that when the cost of AI is lower than human labor and the investment ratio is sufficient (e.g., over 20% of GDP), the rapid accumulation of AI agents creates a positive feedback loop: automation boosts output → reinvestment in more AI → a snowballing expansion of productivity.

Research by Epoch AI and others indicates that if the annual cost of an AI system falls below $15k to complete tasks equivalent to human work, and hardware efficiency continues to improve, the global economic growth rate could surpass 30%. In the optimistic scenario of the World Bank's 2026 report, AI-driven productivity gains could bring global growth in the 2030s back to or even above the highs of the 2000s. The IMF also believes that AI investment has already significantly contributed to U.S. GDP growth in 2026 and could add an additional 0.1 to 0.8 percentage points to global growth in the medium term.

Key mechanisms include: task automation, productivity improvements in individual tasks, and AI-accelerated R&D itself (recursive self-improvement). Institutions like Morgan Stanley predict global growth of about 3.2% in 2026, with AI capital expenditure as a major support.

Real-World Investment Wave and Energy Infrastructure Bottlenecks

By 2026, AI investment has moved from concept to large-scale deployment. Data center electricity consumption has become the most intuitive indicator. IEA data shows that global data centers consumed approximately 485 TWh of electricity in 2025, expected to double to 950 TWh by 2030, accounting for about 3% of global electricity. U.S. data center electricity demand may rise from 80 GW to 150 GW between 2025 and 2028.

McKinsey estimates that by 2030, AI-related data center infrastructure will require $5.2 trillion in investment, with technology hardware accounting for 60%. Giants like Alphabet, Amazon, and Meta plan to invest over $350 billion in 2025-2026. On the energy side, renewable energy purchase agreements (PPAs) have surged, but grid bottlenecks, land constraints, and water consumption (some large data centers use millions of gallons of water per day) pose real constraints.

While these investments drive short-term growth, they may also trigger capital shortages and rising interest rates. In an explosive growth scenario, high return expectations reduce the willingness to save, while infrastructure needs push up borrowing costs, leading to higher long-term bond yields. This could in turn suppress asset prices, creating a complex dynamic equilibrium.

Employment Transformation: Automation Risks and the Cost Disease Effect

The impact of AI on employment is not simply about substitution. Jobs in areas subject to automation will face rapid disruption, but there is still room in non-automated fields (such as occupations requiring physical dexterity or complex interpersonal interactions, like plumbers). Historical experience shows that sectors with rapid productivity gains tend to raise overall wage levels through the "Baumol Cost Disease," causing wages in low-productivity sectors to rise as well, providing a buffer for displaced workers.

Experts predict that even under a "fast AI progress" scenario, labor force participation may decline by 2050, but GDP growth could accelerate to about 3.5-4%. The Wharton model is more conservative: by 2035, AI will raise productivity and GDP levels by 1.5%, and by 2075 by 3.7%. China stands out particularly in robotics and embodied AI, combining manufacturing hardware strength with AI software planning, likely to take a leading position in supply chain and physical industry integration.

Capital Market Implications: Valuation vs. Macro Signals

High valuations in Silicon Valley reflect bets on the long-term dominance of AI companies, but in the money markets, explosive growth has not yet been fully priced in. Long-term bond yields are a key indicator: if they rise significantly, it suggests the market believes in an overall economic "explosion"; if only AI company valuations are high, it is more likely a continuation of a normal growth cycle driven by specific technologies.

Compared to the internet bubble, the difference with AI lies in its ability to accelerate the knowledge frontier. If AI can generate research ideas and break through scientific bottlenecks, its impact on long-term living standards will far exceed that of the internet era. Stanford's 2026 AI Index shows that AI adoption is accelerating historically, with businesses and consumers already deriving substantial value from it.

Policy, Regulation, and Global Imbalances

AI growth potential is unevenly distributed across countries. Advanced economies, with their digital infrastructure and human capital, are better positioned to capture the dividends, while emerging markets need to bridge the digital divide. At forums like APEC, China emphasizes the deep integration of AI with physical industries and the development of robotics, actively positioning itself. The world needs to balance innovation incentives with regulation: data privacy, ethical standards, restrictions on robot deployment, etc., can all become bottlenecks.

Fiscal policy should focus on retraining, infrastructure, and R&D subsidies. Combining energy security with AI investment could become a new growth engine.

Outlook: A Future of Cautious Optimism

AI has the potential to significantly boost global productivity and economic growth, but the explosive scenario depends on multiple conditions such as self-improvement, cost reduction, and breakthrough of bottlenecks. Under the baseline forecast, AI will contribute steady growth momentum from 2026 to 2030; the optimistic scenario could bring historically high growth; the pessimistic scenario might be just another wave of technological advancement, with significant disruption but limited overall output gains.

Policymakers, businesses, and investors should closely monitor actual productivity data, progress in energy supply, labor market adjustments, and bond market signals. The AI era is not an inevitable utopia but a window of opportunity that needs to be actively shaped. Only through technology governance, talent investment, and international cooperation can its inclusive potential be maximized.