Mid-January saw the Prime Minister unveil the Government's new ' AI Opportunities Action Plan ' — a moment that could become one of the most defining for the UK’s AI, data centre, and research communities since the data-powered response to the Covid-19 pandemic.
It’s an excellent paper, and many of Matt Clifford’s suggestions have been met with a resounding welcome by the UK’s digital infrastructure community. While the plan presents a well thought-out and pragmatic approach to ensure the UK remains a prominent leader in AI, the country must now move quickly to capitalise on the economic opportunity it presents.
Subscribe today for free
Time is of the essence
At a fundamental level, the next five years will be critical for the UK to further establish its AI credentials and lay the infrastructure and governance to realise the opportunities surrounding AI. As the Prime Minister clearly stated, any delay will hinder the UK’s chances, and more so as other countries begin to accelerate their development in the field.
According to the US International Trade Administration, “the UK AI market is worth more than £16.6 billion (USD$20.4 billion), and is expected to grow to £788.41 billion (USD$971 billion) by 2035." Much of this growth will only be catalysed via direct partnership with the data centre industry, and its collective ability to provide the secure, scalable and sustainable compute foundations essential to realise the government’s AI ambitions.
China and the US — especially given President Trump’s announcement of a $500 billion joint AI venture between OpenAI, Oracle, and SoftBank — will always be the global AI heavyweights, but the UK currently holds a key position as the world’s third-largest AI market, driven largely by the outstanding intellectual, academic, and entrepreneurial talent we possess in this country. That position, however, is increasingly coming under threat from countries such as South Korea and those in the Middle East, who are doubling down on their AI infrastructure investments - providing computational GPU power at a game-changing scale.
Given these country’s ‘sledgehammer’ approach to cracking the AI nut, it’s vital we accelerate our own investments in critical infrastructure and deploy the much-needed high performance computing (HPC) resources to keep our country’s leading startups, scaleups and researchers here, along with providing the resources for enterprise to harness the benefits of AI alongside the UK population. This will also avoid talent drain and an exodus of intellectual property to vast and easily accessible ‘AI server farms’ abroad.
The same investments will also go a long way towards helping attract and retain the world-leading tech companies pioneering AI such as Microsoft and OpenAI, Google, Oracle, CoreWeave, and Ori Cloud – all of whom have established, and will continue to need, a strong presence in the UK and Europe.
A legacy of innovation
Crucially, the Action Plan correctly states that the UK already has all the raw ingredients required to achieve the government’s ambitions. The UK has a research legacy steeped in innovation from some of the world’s tech pioneers – Alan Turing, Sir Charles Kao, and Geoffrey Hinton, to name but three – but it also has decades of leadership deploying the latest in HPC, AI, and supercomputers.
In 2020, Nvidia’s original Cambridge-1 supercomputer, deployed at our Harlow campus was the fastest and most powerful the UK had ever seen, and provided UK-based researchers with the computational horsepower to identify breakthrough advances in healthcare. This trailblazing project, delivered at the height of the pandemic, showed how investing in data centres and computational infrastructure to support research could create huge advances for society.
It’s imperative that UK data centres, and the renewable energy facilities powering them, are co-developed simultaneously, and AI compute is both accessible and distributed across the country. This will not only aid our entrepreneurial schemes and researchers but will help our startup and scaleup culture to thrive, importantly, in all regions of the country.
Moreover, a diversification of compute will increase both national security and the resilience of our country's data centre sector outside of the crowded West London cluster. To make this a reality, the industry will need to deploy tens of billions in capital investment across the government’s future AI Growth Zones (AIGZs).
AI Growth Zones
Following the designation of data centres to Critical National Infrastructure last year, the establishment of AIGZs is arguably the most important aspect of the Action Plan and a welcome sign of joined-up thinking between government and industry. Three key issues, however, remain fundamental to their success – planning approvals and appropriate land, sustainability, and the energy provision at scale to meet the demands of AI infrastructure.
In the first instance, planning from national, to regional, to local level must be on the same page. Disjointed planning processes can cause significant delays to developments, and we don’t have the luxury of time in the AI race. It typically takes nine months to prepare a data centre planning application before submission, and thereafter, undertake the process of educating the local authority about what a data centre is, and the benefits it can bring.
Streamlining this process with one consistent approach is key, and we at Kao Data believe the government must create central guidance as part of a National Policy Framework to advise local authorities on planning, with a specific use class for data centres. This type of guidance already exists in France, Germany, and the Netherlands and it’s time the UK connected the policy dots.
Providing tax relief could also encourage a surge of private sector investments. One suggestion could be those local authorities retain full business rates for data centres in their jurisdiction, incentivising them to support developments while simultaneously reaping the economic benefits. This approach could ensure local authorities seek to attract and collaborate with data centres, rather than dissuade them.
From a sustainability perspective, local authorities within AIGZs need to proactively identify suitable brownfield sites for redevelopment, thereby mitigating the environmental impact of unnecessarily building on greenbelt. Today there are enough UK brownfield sites to safely accommodate the data centre capacity we need, without incurring the wrath of local communities by developing on greenbelt.
Our £350 million data centre in Greater Manchester (GM) is an excellent example of how regional government leaders like GM Mayor, Andy Burnham, can work together with industry to redevelop ex-industrial sites for the betterment of the region’s AI, cloud and enterprise ecosystems. With all of its positives for a future AIGZ in the north of England, I can’t think of a better location than Greater Manchester.
Power provision
With the energy demands of AI models at an all-time high, and grid constraints slowing UK data centre growth, access to renewable power remains a key hurdle for the government to tackle. We only have to look at the situation in Dublin, Amsterdam, and West London to realise what happens when demand outstrips supply. To solve this, we believe it’s vital that adjacent renewable energy developments are underpinned commercially by data centre buildouts and that the two are co-developed.
According to the Institute of Economic Affairs, “the UK is woefully uncompetitive in both industrial and domestic electricity markets with the highest prices among the 28 countries covered by the IEA.”
UK prices are also predicted to be 2.8 times those of the US. This weakens the UK’s competitiveness and incentivises data centre operators to deploy in locations like the Nordics and Spain, where renewable power is cheap and abundant.
A host of measures to provision clean power must now be sought as part of the new AIGZs. While in the short term, operators are building data centres alongside power cables that feed out to off-shore wind farms, solar arrays, and using private wire to connect to power plants, a longer term and more ambitious outlook would be to use the AI opportunity as the catalyst to explore a more active role for Small Modular Reactors (SMRs), enabling the UK harness the benefit of AI through nuclear energy.
Microsoft is already exploring this opportunity in Pennsylvania, US, where the Three Mile Island nuclear plant will be restarted as part of a new energy-sharing agreement to power its data centres directly for AI. Looking forward, making nuclear power cheaper, more accessible, and easier to build is one of the solutions to solve UK energy issues, but it will require substantial investment and will take both considerable time and effort. Therefore, using shorter-term measures is a necessary and interim solution, while SMRs could be an enabler of future AI growth.
Get the foundations right
In the final part of the Action Plan, the government states it ‘will need to commit to securing the physical infrastructure and human capital that will underpin all future AI developments.’
We at Kao Data firmly agree that collaboration between government and industry is key to ensure that society can benefit from the opportunities offered by AI and we are delighted to see the measures set out by Matt Clifford. Furthermore, we agree that business-as-usual is nolonger an option, and the government must be prepared to absorb some risk in the context of uncertainty.
The future of the AI Opportunities Action Plan and the Governments economic ambitions can only be delivered if we get the data centre foundations right. To ensure the UK continues to be an “AI maker” the country needs both industrial-scale leadership, and data centres, engineered for AI.
RELATED STORIES
UK Gov launches ambitious plan to make UK 'irresistible' to AI firms
Kao Data begins work on £350m Manchester data centre
Kao Data taps CBRE to manage data centre facilities for European expansion
UK Gov grants data centres national priority status to speed up planning permissions
