The Great Divergence, according to the White House

I just got a fascinating document sent over from the White House. Bearing in mind, the White House is the American civil service and not directly communicating Donald Trump’s opinions, thank goodness, their reports are often really insightful. This one really hooked me as it is all about AI and the Great Divergence.

The Great Divergence refers to the impact of the Industrial Revolution, where Western economies rocketed in prosperity and growth at the expense of China and India. The White House believes that AI is going to be a similar revolution. That’s massive impact!

Purely because it is easier these days to do this, I didn’t read the full 27 pages of the report in-depth but just got myself a short summary … through AI!

Abstract:

The report argues that artificial intelligence could trigger a new “Great Divergence” in global economic growth, much like the Industrial Revolution, with the United States positioned to lead through heavy AI investment, rapid innovation, and supportive policy. It highlights AI’s accelerating impact on productivity, GDP growth, and business transformation, driven by scaling laws, falling costs, and soaring investment in compute and infrastructure. While AI is already reshaping labour markets and firm behaviour, its long-term economic effects remain uncertain, requiring close monitoring of adoption, performance, and investment trends. Overall, the report concludes that AI is becoming a foundational economic force, and countries that invest aggressively in AI capabilities and infrastructure are likely to gain lasting competitive advantage.

Introduction to AI and Economic Divergence

The report discusses the potential for artificial intelligence (AI) to create a second Great Divergence in global economies, similar to the Industrial Revolution.

• The Industrial Revolution led to accelerated growth in industrializing nations.
• The Trump administration is focusing on American AI dominance through innovation and deregulation.
• The report aims to analyze current empirical data on AI’s economic impact.

Future Outlook on AI’s Economic Impact

AI is expected to significantly boost productivity and economic growth in the U.S., but the effects will vary and require ongoing monitoring.

• Generative AI, particularly large language models (LLMs), is anticipated to transform economic structures.
• Long-term projections for GDP growth are being re-evaluated due to AI integration.
• Short-term volatility and uncertainty in AI investments are acknowledged.

Understanding Artificial Intelligence Terminology

The report clarifies key AI concepts and distinctions between types of intelligence.

• AI encompasses various systems, from chess-playing computers to generative AI like ChatGPT.
• Current AI is classified as “specialized” intelligence, while “artificial general intelligence” (AGI) and “artificial superintelligence” (ASI) are theoretical future developments.
• AGI would perform all human tasks, while ASI would surpass human intelligence.

AI’s Projected Impact on GDP

Estimates of AI’s impact on GDP vary widely, reflecting uncertainty in economic characteristics.

• AI could increase GDP by 1% to over 45%, with mid-range estimates from various studies suggesting increases of 1.8% to 15%.
• In the first half of 2025, AI-related investment contributed an annualized GDP growth of 1.3%.
• Historical comparisons indicate that AI’s potential impact could mirror past technological revolutions.

International Economic Growth Trends

Different countries are experiencing varied growth trajectories, with the U.S. showing accelerating growth compared to Europe and China.

• The U.S. has a stronger productivity growth rate, while China’s growth is slowing to emerging market levels.
• The EU’s share of world GDP has decreased from 27% in 1980 to 14% in 2025.
• Cumulative private AI investment in the U.S. exceeded $470 billion, compared to $50 billion in the EU.

AI’s Impact on Labor Market Dynamics

The employment effects of AI are mixed, with some sectors experiencing job losses while others see growth. ​

• Employment is declining for early-career workers in AI-exposed fields, but overall unemployment remains low at 4.4%. ​
• Jevons’ Paradox suggests that increased efficiency from AI could lead to greater overall labor demand. ​
• Historical trends indicate that disruptive technologies typically create new job opportunities.

Key Metrics for Tracking AI Progress

Monitoring AI’s rapid advancements is crucial for understanding its economic implications.

• AI training compute has increased by about 4-fold per year since 2010. ​
• Metrics such as investment levels, AI capabilities, and adoption rates serve as indicators of AI’s economic impact. ​
• These metrics are interrelated, with investment driving advancements in AI performance and usage.

Scaling Laws and AI Model Performance

AI model performance improves with increased parameters, training data, and computational power, following empirical scaling laws. ​

• Scaling laws have led to over a billion-fold increase in compute used for training AI models since 2012. ​
• The amount of compute used for training models like Grok 4 and GPT-4 has reached unprecedented levels, costing hundreds of millions to train.
• Global corporate AI investment reached $252 billion in 2024, with generative AI alone accounting for $34 billion, a 19% year-over-year increase. ​

Investment Trends in AI Infrastructure

Investment in AI infrastructure has surged, driven by the growing demand for AI technology. ​

• Investment in information processing equipment and software in the U.S. grew at an annual rate of 28% in the first half of 2025. ​
• This sector now comprises one-quarter of all U.S. investment, indicating AI’s significant impact on GDP growth. ​
• The costs to train AI models have increased at an average rate of 2.4x per year, with cloud compute costs growing at 2.5x per year. ​

Performance Metrics of AI Models

Continuous investment has led to improved performance metrics for AI models, including benchmark scores and cost efficiency. ​

• AI models are achieving near-perfect scores on older benchmarks, with performance on coding benchmarks like SWE-bench jumping from 4% to 72% from 2023 to 2024. ​
• The length of tasks AI can complete has doubled every 7 months over the past 6 years, indicating improved capabilities. ​
• The cost per token for AI models is decreasing significantly, with prices falling by at least 9x to 900x per year.

Adoption and Usage of AI Technologies

The adoption of AI technologies is rapidly increasing across various sectors, reflecting improved capabilities and reduced costs. ​

• The share of U.S. firms using AI in production rose from 4% in 2023 to about 10% in September 2025. ​
• Paid subscriptions to AI services among companies increased from 7% in January 2023 to 45% in 2025. ​
• Approximately 40% of U.S. workers are now using generative AI at work, showcasing widespread integration into the workforce. ​

Global AI Investment Comparisons

The U.S. leads in AI investment, performance, and adoption metrics compared to other countries. ​

• In 2024, the U.S. had $109 billion in private AI investment, significantly higher than China’s $9 billion. ​
• The U.S. accounts for about 75% of reported venture funding in generative AI startups. ​
• Countries like Israel and South Korea are also investing heavily in AI, with Israel’s R&D spending at 6% of GDP, the highest globally. ​

The Trump Administration’s AI Strategy

The Trump administration is implementing policies to enhance the U.S. position in AI through investment, deregulation, and infrastructure development.

• The One Big Beautiful Bill Act aims to increase investment in IT infrastructure and data centres, predicting a 7-10% increase in investment.
• Trade agreements are securing trillions in foreign investment, particularly in AI sectors.
• The AI Action Plan focuses on rapid data centre construction, innovation acceleration, and maintaining free speech in AI development. ​

If you want the full report, you can download it here or have a look through below:

 

Postscript: beyond the AI hype, a 2025 MIT report reckons that most AI fails to do the job (report here)