Breaking Down Global Government Spending on AI

Governments are scrambling to stay ahead of the AI tsunami, and for good reason. Like any other useful technology, AI presents a gigantic economic opportunity for governments worldwide.

In fact, PricewaterhouseCoopers (PwC) has estimated that AI could contribute up to $15.7 trillion to the global economy in 2030 – which would be more than the current economic output of both China and India combined. PwC has also provided a breakdown of where this money is likely to come from. The organizations stated that about $6.6 trillion will come from increased productivity, while an additional $9.1 trillion will likely come from “consumption-side effects.”

To learn more about how governments spend on AI and what we can expect from government AI spending in the next few years, let’s dive into some specific numbers.

The Opportunity AI Presents

Clearly, AI is going to have a massive effect on the economy as a whole. As such, governments are looking to invest in ways that give their citizens a leg-up in future economic endeavors. However, government officials also have their eyes on solving government-specific problems.

We’ll go into more detail about each government’s investments in sectors like healthcare and national security in future articles, but one of the more interesting specific benefits of AI in government is helping overcome bureaucracy and regulation.

One might argue that a government’s sole reason for existence is to uphold laws, but many of these regulations can be difficult to navigate – both for average citizens and experienced government workers.

For the former, look at the “Ask Jamie” virtual assistant in Singapore. This AI tool is meant to help citizens and businesses navigate government-provided services across nearly 70 government agencies. “Ask Jamie” is powered through both chat and voice and is meant to make life easier for Singaporean citizens.

Singapore’s government has made past commitments to staying on the cutting edge of technology.

For the latter, we can look to a recent interview I conducted with Argonne National Laboratory’s own Rick Stevens. While we discussed a lot of topics relating to Argonne’s AI for Energy report, one of the more interesting aspects of our conversation was about nuclear reactors.

As you might imagine, nuclear reactors are some of the most complicated systems humans have ever created. While they are magnificent achievements of innovation that will help wean us off fossil fuels, nuclear reactors can also be extremely dangerous when implemented incorrectly.

The AI for Energy report stated that building advanced nuclear reactors in the U.S. is a “slow, expensive, and convoluted regulatory process.” While the report states that obtaining a construction permit and operating license for a new reactor in the U.S. generally takes five years, the process can sometimes stretch into multiple decades.

By training on datasets of scientific literature, technical documents, and operational data, multi-modal LLMs could help streamline and expedite the nuclear regulatory licensing and compliance process. Considering much of government work is finding ways to cut through bureaucratic red tape to actually accomplish something, AI stands to completely change how our governments operate.

Regional Investment Strategies

Although AI is a widely used tool at this point, its implementation will differ based on region and local government. In later articles, we’ll break down global spending by region much more granularly.

Here, we’ll give a quick overview of what’s happening around the world with government AI spending.

China

  • In July 2017, China’s State Council announced the New Generation Artifiical Intelligence Development Plan. Since then, total Chinese national and local government spending to implement the plan has not been publicly disclosed.
  • By 2022, the Chinese government had reportedly created 2,107 guidance funds with a registered target size of $1.86 trillion. However, by 2023, a report from Zero2IPO stated that these funds had only raised a total of $940 billion.
  • One of the specific regional numbers we do have is from 2018, where Shanghai announced it would launch a fund of about 100 billion yuan (about $14.6 billion at the time) for developing China’s AI industry.

European Union

  • Like China, the European Union announced a national plan for AI investment called the AI Innovation Strategy. It includes a public and private investment package of around €4 billion through 2027 dedicated specifically to generative AI.
  • The AI Innovation strategy calls for a variety of measures, beginning with the intention to create “AI Factories” across the European Union that will bring together supercomputing infrastructure and human resources to further develop AI applications
  • Additionally, the Commission intends to make data available through the development of “Common European Data Spaces.” The objective here is to improve the availability of and access to high-quality data for start-ups and other innovation organizations to train AI systems, models, and applications.
  • On top of the AI Innovation Strategy, the European Union has also established the AI Act to establish a comprehensive legal framework for AI. While not specifically aimed at innovation and growth, the AI Act is meant to foster trustworthy AI that respects fundamental rights and ethical principles.

United States

  • Like the European Union and China, the United States also has a plan in place concerning AI in the form of the U.S. National AI R&D Strategic Plan. Updated in 2023, it outlines the federal government’s roadmap for AI research and development. The United States also has the National AI Initiative Act
  • In 2022 fiscal year, Federal government spending on AI hit $3.3 billion, which is a 2.5 times increase over 2017’s $1.3 billion.
  • The overall United States Federal IT Budget for 2025 is projected to be $75.13 billion, with a heavy focus on cybersecurity and AI.
  • The Department of Defense has been a major driver of AI spending, with AI-related federal contracts increasing by almost 1,200% from $355 million in August 2022 to $4.6 billion in August 2023.
  • The United States will also be the largest market for AI-centric systems, accounting for more than 50% of all AI spending worldwide.

Notable Mentions

  • Japan: Spearheaded by the Ministry of Economy, Trade, and Industry, Japan is working with Nvidia as well as other Japanese companies to unlock the economic potential of AI in the country. Japan has allocated approximately ¥114.6 billion ($740 million) to subsidize the AI computing industry in the country.
  • India: The Indian government has recently announced the IndiaAi Mission initiative to advance the country’s AI ecosystem. Out of the investment of 74 billion Indian rupee (US$1.25 billion), around 45 billion Indian rupee ($543 million) will be used to build computer infrastructure while 20 billion Indian rupee ($241 million) will be used to finance startups.
  • South Korea: The South Korean government plans to invest 9.4 trillion won ($6.94 billion) in AI by 2027. This money is meant to help the country retain an edge in the semiconductor industry and develop AI chips, such as artificial neural processing units and next-generation high-bandwidth memory chips.

This may not represent a global roundup of all investments in AI, but the amount of money the major players are putting up toward AI development is certainly noteworthy.

Challenges and Barriers

While governments are seeing the benefits of investing in AI, there will be some challenges to overcome on both national and international scales. To begin, some social issues will need to be solved over time.

There is a notable AI skills shortage in most countries, and many of the existing workforce may be resistant to adopting new AI technologies. While both of these problems can be assisted by government-funded educational efforts, there are more specific problems for nations to address.

A larger problem that will require a lot of money to address is that many of the legacy systems that government agencies use aren’t designed to work with AI/ML implementations. Solving this will involve modernizing data, networks, the cloud, and cybersecurity capabilities on an enormous scale.

Finally, the overall cost of AI infrastructure will hamper a government’s ability to implement these tools quickly. A recent poll found that 55% of respondents reported that the most significant barrier to the adoption of AI-enabled tools was cost. The various hardware required for AI work has exploded in price recently, and those costs aren’t going to go down anytime soon. Any investment in AI on a governmental scale will demand a hefty upfront cost that some nations simply cannot afford on their own.