Recent headlines have put the spotlight on Big Tech’s investment trends in data centers. While the gang of five has already spent $155 billion on data center expansion since January, the total annual amount for this year is expected to reach $365 billion. That is more than the nominal 2023 GDP of over 150 countries, including Colombia, Czechia, Chile, New Zealand, and Egypt, for example. The massive amounts of capital these companies have at their fingertips are indisputable. That should serve as the best qualifier for the “big” adjective typically used to band them under the same canopy.
The news reports also indicate that intense competition among them—and with others, I should add—propelled by the accelerated development of AI, is driving such financial decisions. I am uncertain about what the advocates of intellectual monopoly capitalism or technofeudalism will have to say about this. They will probably try to convince me that competition is a 19th-century phenomenon.
At the policy and strategic levels, advanced capitalist countries are also taking swift action. While national AI policies have been in place for almost a decade, renewed efforts have shifted from responsible and ethical AI to a focus on data centers as the primary infrastructural driver for AI development and global supremacy. The recent US AI Action Plan (AIAP, PDF) is a great example. It piggybacks on the Data Center Infrastructure Acceleration Executive Order, which essentially eliminates most regulatory and environmental constraints for deploying the new digital factories nationwide. AIAP has over 90 action items. Almost a third of them focus on strengthening data center capacity, including energy and workforce requirements, and linking these efforts to semiconductor manufacturing, cybersecurity, and military use.
China’s recent Global AI Governance Action Plan (GAIGAP) espouses a significantly different approach, with only 13 action items. Two of them are related to data centers and AI infrastructure. The rest are apparently a call for a global, multilateral approach to ensure the benefits of AI spread worldwide while its risks and challenges are addressed jointly. Surely, both plans are permeated by geopolitics and national ambitions to dominate that space. However, GAIGAP sounds like something the EU could have drafted, paradoxically. By the same token, AIAP is shamelessly ideological and openly imperial.
The UK’s approach is somewhere between AIAP and GAIGAP. Of course, the UK is no slouch in this regard. According to the 2024 Stanford Global AI Vibrancy Tool, the UK is only behind the US and China, the top leaders. In the past six months, it has delivered three interconnected policy and strategy documents. The AI Opportunities Action Plan was unveiled in January. The Modern Industrial Strategy 2025 was completed in June. Finally, the UK Compute Roadmap was published last month.
The AI plan has three components. The first one lays the foundational enablers for AI growth in the country. This includes data centers, data assets, talent, and AI scientific development, as well as “pro-innovation” regulation. The second embraces the moonshot or missions approach developed by Mazzucato. It thus gives the state a more prominent role in implementation, albeit one that is always worked out in collaboration with the private sector and other partners. The last one depicts the ultimate long-term goal of the action plan: to create homegrown AI capabilities, infrastructure, and businesses. In this light, it proposes creating a UK Sovereign AI unit to ensure that this goal is achieved in practice.
The industrial strategy is much broader in scope, as expected. It has four pillars. 1. Promote business development. 2. Support regional development via productive clusters. 3. Prioritize economic sectors with the higher potential. 4. Foster partnerships between the business sector and a more agile, capable, and stronger state. It sees AI as one of the critical enablers. The strategy also supports calls for sovereignty that go beyond AI. Here, we once again see the use of the missions approach and repeated calls for sovereignty in strategic sectors. Creating UK businesses that can compete with Big Tech is one of the clarion calls.
The UK Compute Roadmap is the government’s response to the rapidly changing global environment. Indeed, digital infrastructure, in the form of data centers, has become a strategic enabler. It is attracting substantial private investments, as mentioned above, where Big Tech and others holding massive capital amounts can spend freely to outdo the competition. At the same time, AI is now introducing dramatic changes in supply chains and sectoral organization while fostering new alliances across the board.
While the UK is indeed a leader in digital technologies and AI, the strategy highlights the fact that the country’s infrastructural ecosystem is fragmented and uncoordinated. That could have pernicious consequences, potentially impacting economic growth and current living standards. The roadmap addresses such challenges without ever overlooking the fact that digital infrastructure is a means to an end. The real targets are fostering scientific advancement, promoting industrial competitiveness, and enhancing public service delivery.
The roadmap sets the long-term vision and carves the path for creating world-class computing infrastructure. It centers on four core pillars encompassing ten action items. It also defines concrete funding commitments and outlines the institutional transformations required for successful implementation. The last pillar could be positioned as the long-term outcome of the roadmap: creating sovereign, secure, and sustainable capabilities for designing and building compute technologies. That involves backing British companies that can compete globally, with the open and targeted support of the Sovereign AI unit. The other three pillars focus on building state-of-the-art compute infrastructure, creating a modern public compute ecosystem, and powering innovation in both the public and private sectors.
Institutionally, the roadmap builds on existing initiatives such as the AI Research Resource (AIRR), launched as part of the AI action plan, and the National Supercomputer Centers (NSCs). It also enhances the existing AI Growth Zones, linking them to regional development in sync with the industrial strategy. Finally, the Sovereign AI Unit is slated to receive an injection of £500 million and increase its focus on building homegrown capacities in emerging areas of AI.
Now we can proceed to compare the three approaches. AIAP is specifically tailored to the ambitions of the current US administration. Not surprisingly, it has reversed most of the AI policy proposals from the previous administration. In this light, no other country could adopt a similar approach, except for China. Nevertheless, the latter’s approach seems relatively subdued at the global level, at least on paper. GAIGAP is, after all, an international AI governance proposal first and foremost, where a leading AI country is positioning itself as the lead. Again, no other government could do that, except for the US. On the other hand, the UK perspective is more inward-looking. It has an upbeat approach towards digital sovereignty, backed by critical financial resources and responsive institutional structures and governance mechanisms.
From the perspective of the Global South, the UK is then the most relevant at the policy level. In practice, developing and emerging economies should be fully aware of the substantial differences between them and a former imperial power, however.
The high degree of policy coherence and coordination in the UK’s computing, AI and national development aspirations is one critical area that developing countries should note and prioritize. More often than not, policy fragmentation and sheer competition among public institutions prevail in the latter. Nonetheless, cohesive policy development is a function of the level of state capacity, which, in most nations of the Global South, is comparatively weak. Consequently, enhancing state capacities should also be part of the overall policy equation. Digital technologies can undoubtedly play a crucial role here.
That is closely linked to the prominent role the UK has given to the state in the overall policy and implementation process. Pushed by international financial organizations, developing economies typically limit the state’s role to a minimum set of activities, explicitly avoiding any direct involvement in policy implementation. The UK roadmap takes precisely the opposite approach, as it is convinced that without direct state involvement, the country will miss the AI boat. The strategy to empower the state is rooted in the mission or moonshot approach. This perspective openly acknowledges the critical role of state involvement in ensuring that the benefits of speed-of-light technological innovation are directed towards overall human development and improvements in living standards across the board.
The mission approach also calls for strengthening links and partnerships between government, businesses, and academia to promote innovation and its inclusive diffusion. High-impact priority sectors are typically prioritized as the core targets of long-term joint interventions. This is an essential hint for developing economies, as they first need to identify priority sectors to then design and implement technology policies that support their growth and international competitiveness. Indeed, data centers are at the core of this. At the same time, they should not be positioned as the core development outcome.
Enhancing institutional capacities is another critical theme, closely related to augmenting state capacity. The various UK policies and action plans clearly identify the institutional instances that must be in place to ensure the expected long-term outcomes are achieved. Moreover, the UK Compute Roadmap and the AI Opportunities Action Plan were developed by the Department of Science, Innovation, and Technology (DSIT), which has a mandate that is legally and substantially different from that of ICT-focused institutions and entities. DSIT, in turn, supports UK Research and Innovation (UKRI), an independent research entity that works similarly to the US National Science Foundation (NSF). UKRI’s core goal is to boost research and knowledge sharing among higher education institutions and support Innovate UK. The latter is the national innovation agency responsible for nurturing business-led innovation across all sectors. All these entities must work in sync to guarantee the successful completion of the various missions under the radar.
Finally, linking AI and science, technology, and innovation policies (STI) is crucial, as explicitly articulated by the UK Compute Roadmap. Harnessing digital infrastructure as the foundational enabler for the country’s overall future is essential. However, an approach entirely centered on data centers alone will not suffice.
Raul
