Listen to an audio version of this article (generated using Gemini).
Winning the Race
The United States has launched a new national strategy called "Winning the Race: America’s AI Action Plan", developed during the early days of President Donald Trump’s second term and published by The White House this month. The plan outlines more than 90 policy actions to be taken across the U.S. federal government in the short term, its main aim being for the United States to achieve and maintain global leadership in artificial intelligence, which the administration views as key to future economic success, job creation, scientific advancement, and national security. The initiative is built on the belief that succeeding in this global AI race will lead to a new era of prosperity for Americans.
At the heart of the plan are three major focus areas:
- Accelerating AI Innovation.
- Building American AI Infrastructure.
- Leading in International AI Diplomacy and Security.
Each of these pillars supports a different aspect of the overall strategy. The first pillar is about boosting innovation by encouraging the private sector to create advanced AI systems and apply them in ways that drive growth and discovery. The second pillar focuses on improving the country’s energy and technical infrastructure to meet the growing demands of AI technologies, especially in light of stagnant capacity since the 1970s and competition from China’s expanding capabilities. The third pillar concentrates on promoting American-made AI technologies and standards around the world, strengthening alliances, and ensuring that rivals do not benefit from American innovations.
Source: ai.gov
Build, Baby, Build
The Plan takes a firm stance on rolling back what Trump and many of those associated with the White House, especially in the tech sector, view as restrictive rules that could slow down innovation. One of its first moves is to reverse President Biden’s Executive Order 14110 on artificial intelligence, aiming to prevent what it calls unnecessary red tape. The administration wants to identify and eliminate federal regulations that make it harder to develop or use AI and plans to withhold federal funds from states that impose particularly strict AI rules. The plan also recognises that sectors like healthcare have been slow to adopt AI and proposes new “regulatory sandboxes” and Centres of Excellence to allow faster testing and introduction of AI tools. In parallel, it seeks to speed up the use of AI technologies in the Department of Defense and Intelligence Community to strengthen national security capabilities.
Alongside its focus on cutting regulation, the plan outlines ways to support Americans and boost innovation. It introduces a “worker-first AI agenda,” which means developing training programmes and education pathways to give people the skills they need to work in the growing AI economy. The administration also aims to pilot fast-track retraining schemes for those at risk of losing jobs due to automation. Investment in next-generation manufacturing - such as autonomous drones and robotics - is another key priority, especially for defence and security. AI-enabled scientific research will also receive a boost, with funding directed at modern laboratories and better access to public scientific datasets. There is a specific interest in sequencing genomes of life on federal lands. Additionally, the plan prioritises more research into understanding and controlling complex AI systems, especially to make them safer and more reliable when used in sensitive contexts like defence.
To this end, the plan includes several infrastructure and workforce initiatives. It proposes faster approval processes for building semiconductor factories, data centres, and energy projects, including updates to environmental rules like the Clean Air and Clean Water Acts. There’s a strong focus on upgrading the U.S. electrical grid, using advanced energy sources such as nuclear fusion and geothermal energy to support growing AI power demands. The government also wants to encourage domestic semiconductor manufacturing, ensuring good returns on CHIPS Act investments while cutting regulatory delays. To keep sensitive data secure, it plans to create new technical standards for military and intelligence data centres. Preparing the workforce is central to this vision, with an emphasis on training electricians, HVAC specialists, and other skilled trades to build and maintain AI infrastructure.
The Response
Much of the plan - although by no means all - is concerend with bread-and-butter infrastructure funding, and probably two-thirds of it could actually have been produced by the Biden administration under different circumstances. Indeed, in those areas The Washington Post is probably correct when it evaluates the plan as "Trump is off to a good start".
Much of the press coverage, however, has frequently been much more ambivalent or even openly hostile, even where US commentators in particular would probably agree with elements in the Plan that counter Chinese influence in international AI governance, such as through tightening controls on advanced AI computing and semiconductor exports, and urging allies to adopt similar protections. Unsurprisingly, the Trump plan aims to safeguard national security while embedding American AI systems and values into the global digital economy which would account for some of the more skeptical international opinion. Among more positive responses, some praise cutting red tape and boosting private-sector innovation, as well as the goal to transform the U.S. into a global AI export leader, helping American companies compete with China's subsidised tech - although as the Council on Foreign Relations points out, there are also plenty of risks here. The strategy maintains continuity with previous administrations in infrastructure and innovation investment while advancing a worker-first AI agenda focused on skill development and retraining. Bipartisan support is expected, particularly for energy initiatives and securing AI innovations from emerging threats.
Critics, however, argue that the AI Action Plan lacks clarity on execution, with many proposals missing assigned lead agencies, deadlines, or funding details. This raises doubts about whether federal departments have the technical capacity to deliver. While the plan recognises the challenges posed by opaque AI systems, its actions on interpretability and control are seen as inadequate. Concerns also arise around the proposal to restrict funding to states with strict AI regulations, which could trigger federal-state tensions reminiscent of past conditional funding efforts. The omission of copyright and content provenance issues is viewed as a significant oversight, given their importance in AI training and outputs. Further criticism targets the plan’s ideological stance, particularly the "Preventing Woke AI” order, which is considered politically motivated and misdirected. Internationally, the emphasis on American AI dominance may alienate allies with their own transparency and digital rights frameworks. Inconsistencies in export controls, especially the Nvidia chip reversal, undermine strategic goals. Environmental groups warn that the push for rapid permitting and broad energy development could worsen ecological harm and raise utility costs. Finally, expanded federal data-sharing powers and involvement of companies like Palantir have sparked fears of increased surveillance and the emergence of a system resembling China’s social credit model.
This article was co-created with AI.