AI’s next era: industrial policy, global competition and the end of regulation-first thinking

April 2025 | SPOTLIGHT | SECTOR ANALYSIS

Financier Worldwide Magazine

April 2025 Issue

This year’s 2025 AI Action Summit in Paris marked an inflection point in the history of artificial intelligence (AI). For years, the focus of governments worldwide was on careful regulation and safe development of AI, marked by such milestones as the EU AI Act. But that time may have ended, to be replaced with a new era shaped by global competition. It began with the name. This year’s event threw out the old title ‘AI Safety Summit’ for ‘AI Action Summit’. At the event, JD Vance, US vice president, delivered a bold speech, criticising European leaders for excessively regulating the AI industry. The event concluded with the publication of a non-binding global commitment to safe and inclusive AI, which the US and the UK refused to sign.

Thus, what began as a push for broad safety regulations has evolved into a race for technological supremacy, with nations prioritising competitive advantage over precautionary measures. This shift reflects growing recognition that AI leadership confers economic and military power, making it too strategic to constrain with heavy-handed rules. The result is a new landscape where industrial policy, infrastructure investment and geopolitical calculations have superseded sweeping regulation as preferred policy tools for AI.

Regulatory retreat

Recent years have seen major economies pulling back from comprehensive AI regulation in favour of approaches that prioritise innovation and competitiveness. The US exemplifies this trend, with the current administration revoking executive orders on AI issued by the previous administration. This reversal reflects concerns that excessive rules might handicap American technology companies facing Chinese competition. Washington’s stance now explicitly prioritises AI leadership and national security over precaution. Among US states, only Colorado has adopted comprehensive AI regulations, but even those do not take effect until 2026. Other states have limited regulations to narrow restrictions addressing specific AI harms caused by certain applications, such as automated decision making in employment and consumer decisions, deepfake media, and biometrics.

The European Union (EU), despite championing the AI Act as the world’s first comprehensive AI law, has recognised that excessive red tape could undermine its tech sector. European leaders have supplemented regulation with pro-innovation measures, attempting to spur private sector investment. One key trend to watch is how the EU interprets the AI Act to see how narrowly regulators and the courts will interpret it. In its most recent guidance, the EU took a notably narrow view on what AI uses were outright prohibited. This dual strategy shows the EU hedging its bets – maintaining rules for trustworthy AI while aggressively supporting industry to remain competitive.

The UK has adopted an explicitly pro-innovation approach seeking to ‘turbocharge’ AI. Instead of creating a dedicated AI regulator or strict laws, the UK issued flexible principles for existing regulators to apply. Officials argue this light-touch stance prevents stifling nascent AI businesses, with the government even framing the EU’s tighter regulations as a competitive opportunity to attract AI firms to British shores.

China’s stance has always differed, focusing on control and innovation simultaneously. Beijing introduced new rules for generative AI and algorithmic oversight, but these largely aim to ensure AI aligns with state censorship and social stability rather than slowing development.

Massive industrial investments

As regulation softens, nations are implementing unprecedented AI industrial policies, pouring vast sums into research, infrastructure and innovation. The EU, alarmed by lagging private investment compared to the US (€9bn versus €62.5bn in 2023), is launching a €200bn AI investment drive to accelerate research and development (R&D) and deployment. This investment aims to counterbalance American and Chinese dominance by giving European companies the capital and resources to compete at scale.

The US has unveiled the Stargate Initiative, a public-private plan of staggering scale – $500bn – aimed at fortifying American AI leadership. Announced in early 2025, Stargate focuses on building next-generation AI infrastructure domestically, including investing roughly $100bn in cutting-edge AI data centres. The initiative involves government and industry partnerships while streamlining regulations to speed up permits for AI projects. But in the US, this is only one of many multibillion-dollar initiatives, with massive investments also announced by Meta ($65bn) and Apple ($500bn).

China has long treated AI as a national priority, pursuing a whole-of-nation strategy that couples enormous investments with state directives. Beijing launched the ‘Eastern Data, Western Computing’ project to construct eight major data centre hubs across its western regions, attracting over 200bn (approximately $28bn) yuan in investment. Simultaneously, China is pouring funding into semiconductor independence to counter US export controls, supporting domestic firms like Huawei in developing homegrown AI chips. Beijing is now preparing a new 1 trillion yuan (approximately $140bn) AI Industry Action Plan to rival the US Stargate project.

The infrastructure battleground

Nations are beginning to recognise that AI dominance requires control of critical infrastructure, namely, semiconductors, cloud computing capacity and energy resources. Advanced AI runs on specialised chips, making them flashpoints of geopolitical tension. In 2023, the US tightened export controls to prevent China from acquiring high-end AI chips, seeking to limit breakthroughs in supercomputing that could aid China. This created a market for modified chips and accelerated China’s drive for semiconductor self-reliance through domestic players.

Cloud computing infrastructure has become equally strategic. The US Stargate plan aims to construct AI mega-data centres exclusively on American soil, while China’s ‘Eastern Data, Western Computing’ project essentially creates a national data centre network inside China. Control over cloud infrastructure translates to influence over AI services globally, with American tech giants currently dominating international markets while Chinese firms lead domestically. European initiatives like GAIA-X aim to reduce reliance on foreign providers, driven by data sovereignty concerns.

The energy demands of AI computing present another challenge. Recent studies show US data centres could consume up to 12 percent of the nation’s power by 2028, up from about 4 percent today. This unprecedented growth has utilities planning new power plants to meet projected demand, though efficiency breakthroughs could alter these calculations. Governments are exploring sustainable solutions, with the US Department of Energy considering small nuclear reactors and advanced geothermal for data centres, while France touts its nuclear plants as carbon-free support for AI computing.

The DeepSeek moment

A dramatic illustration of this new era came with the emergence of DeepSeek R1. In January 2025, a Chinese AI lab called DeepSeek released a powerful open-source AI model that came close to the performance of top models from Western companies. Even more striking, they achieved this with fewer computational resources, suggesting an alternative path to AI leadership through algorithmic innovation rather than just computing scale.

The immediate market reaction saw investors reconsidering assumptions about Western AI dominance. OpenAI responded within days by announcing a free mini-reasoning model, demonstrating how seriously Western firms took the threat. Energy sector stocks fell slightly as stakeholders realised more efficient AI models might temper projected electricity demand growth.

DeepSeek R1 galvanised policy responses globally. The US Stargate Initiative was announced one day after DeepSeek’s release, as if to reassert American resolve. Beijing celebrated the breakthrough as validating its approach, likely influencing its new trillion-yuan AI plan. European leaders saw an opportunity to champion open-source AI as a counterweight to both American corporate control and Chinese state-led models.

Implications for business and investment

For companies and investors, this shift from safety governance to competitive industrial policy carries profound implications. Businesses must prepare for potential AI decoupling between East and West as restrictions on technology sharing increase. This may require separate AI strategies for different markets – using Chinese platforms inside China but Western ones elsewhere. Investors face limitations on capital flows, forcing complex decisions about where to place bets in the evolving ecosystem.

For companies building or operating data centres, the hardware and talent underpinning AI have become potential chokepoints, with geopolitical tensions threatening supply chains. Companies should diversify their sources for critical components like chips and consider contingency plans for cloud services.

Further, while broad AI laws may be in retreat, a patchwork of narrower regulations is emerging, focusing on point-of-application issues where AI touches the real world, such as automated decision-making technologies, anti-deepfake regulation and similar issues. Companies must remain agile, designing systems that can be configured to meet different regional standards. The likely scenario is divergent regulatory regimes – American voluntary frameworks, with narrow state-specific regulatory restrictions, European formal compliance systems and Chinese security-centric rules. Careful governance will still be necessary to develop and deploy AI applications even in this period of AI competition.

This new era of AI governance demands strategic foresight from all stakeholders. Those who master both the governance framework and competitive dynamics of AI will shape the future of this transformative technology. The global AI race is accelerating, and the implications extend far beyond technology to economic power, national security and geopolitical standing.

Matthew Baker is a partner, Parker Hancock is a senior associate and Justin Bryant is an associate at Baker Botts LLP. Mr Baker can be contacted on +1 (415) 291 6213 or by email: matthew.baker@bakerbotts.com. Mr Hancock can be contacted on +1 (713) 229 1196 or by email parker.hancock@bakerbotts.com. Mr Bryant can be contacted on +1 (650) 739 7574 or by email: justin.bryant@bakerbotts.com.

Matthew Baker, Parker Hancock and Justin Bryant

Baker Botts LLP