A Manifesto for the UK in the Third Millenium

April 2024

Introduction

As of writing in the Spring of 2024, the United Kingdom is in a rut. Real GDP per capita continues to tumble and society is fracturing under a vicious cycle of slower growth, higher taxes and an encroaching state. Investment is flat by some measurements and rapidly falling by others. In such a prolonged condition of stagnation, zero-sum thinking has crept into our politics and a broader sense of pessimism evidently grips the national psyche. When the British state cannot even seem to master the most basic responsibilities of government, a manifesto that seeks to position Britain for hundreds if not thousands of years of prosperity may seem more absurd than ambitious. I write this essay not in dismissal of the day-to-day policies needed to bring about change in the short run – for they are deserving of a separate piece in their own right – but as a cause for optimism in the long run. Contrary to the current gloom there is a viable route for the UK to reinvent itself as a new technological hegemon in a universe where human power is orders of magnitude greater. As the capabilities offered by new innovations become ever more powerful, their benefits will become ever more concentrated among fewer key actors. Capturing these is a duty we owe to our descendants. 

Aerospace 

The UK is in a strong position to grow its space capabilities after several underwhelming decades. Government ownership of OneWeb (one of only a handful of companies on earth to offer LEO communication services), industrial backing through the UK Space Agency, and ongoin failure by industry incumbents all offer opportunities. OneWeb was bailed out by the UK government in 2020 at a cost of $500m for a 45% stake. The deal gave the UK a golden share that prioritises the UK to use the service for national security reasons. OneWeb has completed its launch of 650 satellites in LEO for phase 1, which is noticeably less than Starlink at ~5,500, but still impressive in its own right. Competition also comes from Amazon who, through Project Kuiper, have put several test satellites in orbit ahead of an eventual constellation of ~3,300 satellites. 2022 saw OneWeb merge with France’s Eutelsat in hopes of a consolidation to boost each other’s prospects. The UK retains special voting rights and an 11% stake in this new venture, and the company remains based in the UK with preference for UK contracts and launch sites. The economic issue holding back Eutelsat-OneWeb is its extremely expensive ground receiver units which are priced at $10,000 each, which is 10 and 20 times more expensive than Starlink and Kuiper respectively. This gives it no chance in the consumer market, and that has been partially reflected in a share price collapse from €11 per share in 2021 to around €4 today. OneWeb at least sit on an order book of over $1bn for its satellite network, but requires an estimated $6bn extra in funding to complete a full LEO constellation. With full government-backing and an accelerated launch schedule, OneWeb could yet prove a worthwhile venture to back.  

The UKSA launched the Space Based Positioning, Navigation and Timing (PNT) Programme in response to Brexit separating the UK from the EU’s Galileo project for continental autonomy in space-based navigation systems. A fully-expanded OneWeb satellite network could simultaneously provide LEO communication services and PNT. The $6bn cost (plus more for bolting-on PNT technology) of building this out would be expensive in the short term but a strong investment in the longer term. Consider the benefit of having an alternative to the US-controlled GPS in a location tracking industry worth billions of dollars. Firstly, it could be commercialised in direct competition with GPS at a more competitive rate in order to raise revenues. It could also to act as an alternative to when GPS services fail, and avoid a situation where the UK economy, according to official government estimates, would lose £1bn per day. 

Starlink’s dominance sounds insurmountable but its advantage in constellation size is relatively new. In late 2020, Starlink had a similar number of satellites in orbit as OneWeb currently has: around 650. In the three and a half years since, that implies an 84% annual growth rate in the number of Starlink satellites. At this rate, Starlink will reach its FCC-approved goal of 12,000 satellites by the middle of 2025 (although the pace of scaling will likely diminish and so 2026 onwards is a more realistic time frame). OneWeb does not have an equivalent publicly stated target, but from various reports it seems that low-thousands would be sufficient to compete in both PNT and communication services concurrently. At the same rate as Starlink, OneWeb could reach that within two years. The opportunity and timing for the UK here is critical on two accounts. Firstly, for all of Starlink’s impressive growth it has only two million customers and is just a few years old. That is still very nascent relative to the potential of the LEO satellite market; the number of people without good internet access is in the billions and thus the market is still unpenetrated by any company or state. And at a sufficiently competitive cost and quick speed of internet, LEO offerings could compete for the entire global market – even those on conventional broadband packages. Secondly, the UK has the unusual position of having a strategic stake in one of the handful of viable companies in this industry. Within a few years, Starlink and other incumbents would benefit from a network effect and first-mover advantage to completely entrench themselves. But with the industry so young, there is still a window of opportunity currently open as to which firm cements themselves as the market incumbent. The UK could initiate a rapid and government-funded programme to establish economies of scale in LEO infrastructure. That means prices sufficiently competitive to win over the majority consumer, enterprise, and government markets in both communications and PNT (ideally solutions optimised to perform better than GPS and Galileo in latency and accuracy). The PNT services, if superior to GPS, could be heavily monetised in what is a high operating leverage model: the largely fixed cost of its satellite constellation will generate scalable revenue. The cash flow from PNT could in turn subsidise a massive reduction in ground receiver costs for OneWeb LEO internet from $10,000 to <$500 to make it competitive in the consumer market. That means running the OneWeb constellation at a loss or perhaps marginal cost breakeven for at least a decade. But the bull-case would see the UK become the global primary power in the LEO market, establishing itself as the national hegemon in all related infrastructure and systems. The ultimate ambition here is not for OneWeb infrastructure to become a cash cow, but to use its infrastructure and data to pave the way for more ambitious space plans in this solar system and beyond. Outside of OneWeb, there are additional UK specialities in the field of aerospace engineering. One of SpaceX’s first investments in another company was in Surrey Satellite Technology (SSTL) in 2004, which Musk called ‘a high-quality company that is probably the world leader in small satellites’. Airbus Space employs 12,000 people across the UK and works with a network of over 500 SMEs, SSTL among them. Rival Boeing Space is stagnating and the wider company suffers from tumult at its highest levels in early 2024. Its Starline programme with NASA remains delayed by years due to engineering faults and cost overruns, and the wider space division is being scaled back to focus on failings in the aircraft business. This is certainly an industry ripe for disruption by those who will move quickly.  

Beyond the engineering aspect, the UK should take advantage of its sprawling geographical ties to entrench dominance in the logistics of the space industry. The UK and its British Overseas Territories (BOT) offer a wide variety of locations for launches and logistics despite their small sizes. The BOT of St Helena, Ascension and Tristan da Cunha is a string of islands and archipelagos dotted across several thousand kilometres of the south Atlantic. Among these, Ascension is the most northerly and sits at a latitude of just -7.9 degrees below the equator, making it ideal for the usual preference of equatorial launch sites. There is already a surprisingly large amount of aerospace infrastructure on the island. Its Wideawake Airfield was renovated to a length long enough to allow for space shuttle flights, and has gone through phases of being one of the world’s busiest airports during times of war. Given the engineering currently taking place at aerospace startups like Boom and Astro Mechanica, the line between aircraft and rockets may become blurred and developed runway infrastructure may serve as an additional port to space. Elsewhere on Ascension, NASA established a tracking station as well as a Missile Impact Location System that identifies satellites or returning crews that have re-entered from space into the ocean. It is also one of the four ground antennas that form the core of the GPS system – which invites obvious opportunities for OneWeb’s PNT ambitions. NASA even operate a MCAT telescope on the island for the purpose of tracking orbital debris. The European Space Agency also have monitoring facilities there. GCHQ and the NSA jointly run advanced signalling and communication infrastructure from the well positioned island. Even the BBC operate a relay station for their World Service.   

The well positioned geography and established infrastructure by both the UK and key allies makes it an obvious candidate for future launches. The island would become home to the UK and humanity’s primary space port of HMSP Francis Drake, named in tribute of the country’s pioneer in global circumnavigation and exploration. At 34 square miles (~22,000 acres), the island is not large but is sufficiently big enough to cope with a large space port (SpaceX’s Boca Chica site in Texas is just a few hundred acres). If necessary, dredging could be used to extend the size of the island. By building out land onto all waters no more than 30 metres deep, Ascension could be extended by an additional 20% in size, predominantly to its northern and eastern edges. This would allow for up to 20 launchpads, comfortably make it the world’s largest spaceport ahead of Baikonur Cosmodrome in Kazakhstan with 15. The island’s dormant volcano makes it an ideal spot to harness geothermal energy for a community of tens of thousands and the accompanying energy-intensive facilities that allow for space monitoring and launches. Nuclear energy, as outlined later in this essay, could also be a supplement to baseload supply. The island would also need to develop self-sufficiency in petrochemical refinement and storage to ease constraints regarding large fuel consumption. Ascension’s isolation also makes it ideal for a laissez-faire approach to iterated and experimental launches from HMSP Francis Drake. Inhabited land is rightly an unpopular place to risk space debris falling back to earth from a test launch. The nearest dry land to Ascension Island is Saint Helena which is 1,400 kilometres away, and extremely small and with under 5,000 residents. Assuming an easterly direction to any launch from Ascension (as to benefit from the momentum of Earth’s rotation), then the nearest sizeable mainland east of the island is Angola – a full 3,000 kilometres away. This gives a vast area of ocean with which launches can be continuously run and debris cleaned up and retrieved without concerns of impacting human habitation. The UK could consider offering the infrastructure of HMSP Francis Drake as a service to the wider industry, notably for SpaceX, to help finance UK space ambitions and to foster a hub of like-minded companies. Thinking beyond the status quo type of launch infrastructure, there is a growing potential for more work to be operated by sea. That could manifest as using ships to assemble and launch rockets of a significantly large size which are too big for land-based engineering. SpaceX’s use of autonomous drone ships is an early indication for what may become a partially sea-based industry. Ascension’s geography would be clearly conducive to this, especially once its infrastructure has been well developed. 

Successful space policy will be a cornerstone for the UK’s longest-term ambitions. It is paramount that the UK become a major and competitive operator in the payload transport industry and LEO communications market. Maintaining this access to space hardware and data will form the backbone of efforts to unlock extraterrestrial resource extraction, engineer vast projects like terraforming or Dyson spheres, and to ultimately establish a series of space colonies. The non-linear nature of these extraordinary opportunities makes expensive government backing a worthy endeavour. 

Semiconductors and AI  

A combination of geopolitical fallouts and acute shortages during 2021 initiated a reinvention of the semiconductor supply chain. Countries, the US chief among them, poured tens of billions of dollars into the reshoring of manufacturing capacity – and it is proving expensive. The cost of the TSMC fab in Phoenix, Arizona has surged to over $40bn alone. And like all shortages, the scramble for chips during the depths of the pandemic was followed by a glut in 2022 and 2023 in which cyclical NAND and DRAM prices collapsed by ~50% YoY and global semiconductor sales fell by nearly 25%. It is too early to be certain, but we could have a structurally over-supplied market in both non-memory and memory chips over the coming decade as various countries race to expand domestic production without regard for supply and demand dynamics. In such a scenario, it is rational for the UK to not join the rush for onshoring production and instead to step back and absorb the oversupply of semiconductors at lower prices. This is now a sustainable strategy because much of this over supply will come from the US. Of all the West’s 35 fabs under construction, 27 are being built in the US. Even if it wanted to compete, the UK simply does not have pockets as deep as the US, EU, or China to try and woo the likes of TSMC and Samsung to build vast fabs. This critique applies equally to commodified industrial goods like batteries and PV solar panels. Entering such a race would be prohibitively expensive and with little value-add in the long term.  

Where the UK does have a comparative advantage in this field is in asset-light semiconductor software and design. Arm has been the shining light in this field, and as of writing is up 120% on its US IPO just six months prior, reaching a market cap of $140bn. Though on a separate note, its listing on the Nasdaq is indicative of Britain’s decline in our capital markets – another area in need of reform but which is beyond the scope of this manifesto. The instruction sets that Arm licence operate in 99% of smartphones globally, and a growing percentage of PCs as well. Since its founding in 1990 the company has seen over 250 billion of its chips produced. It is a staggering success of where the UK has managed to cement a small but dominant position in the global boom of electronics and, more recently, in AI.  

The UK has two routes to grow its contributions to fabless semiconductor software and design. The first of those is simpler but more expensive, which is to chase industry leaders on the current trajectory of development in conventional semiconductor technologies. That means optimising to create the most powerful designs of CPUs and GPUs through generous but demanding state VC funding of fabless semiconductor start-ups, and grants to leading research centres across the UK. They would be operating under a single entity to strengthen monopsony and monopoly powers on a global market. On the assumption that energy becomes negligibly cheap under the set of nuclear policies outlined below, it would be more worthwhile to attempt to leapfrog market incumbents by focusing efforts on maximising the power of chips versus their efficiency on the assumption power would never again be a bottleneck to either chip performance or the scaling of LLMs. Such a trade-off would be a calculated risk depending on timelines of energy developments, but prioritising raw capabilities with less regard for power consumption would accelerate R&D timelines ahead of competitors. A lot of the advantages enjoyed by market incumbents in the fabless industry are ones that the UK government could summon rapidly. The first being relationships with major foundries. TSMC had revenues of $70bn in 2023. A large government-backed order of say, $7bn, would increase the Taiwanese company’s revenues by 10% and make it the second largest customer behind only Apple. Such an order would automatically secure a strong relationship that is critical to taking advantage of TSMC’s most advanced capabilities and to ensure prioritisation for production during periods of market tightness. This would be strengthened by good bilateral relations with the island as evidenced by their government’s 2023 deal with Eutelset-OneWeb for LEO communication services.   

The second area of semiconductor software worth exploring would be in alternative theories of computing. By virtue of novelty, the cost of R&D for new ideas in say, natural systems of computing and analogue chips, is certainly lower than competing at scale in proven technologies with the world’s largest technology firms. Despite being cheaper, alternative forms of computing are still theoretical and therefore face a high degree of failure as well – at least individually. The UK’s state-backed Advanced Research and Invention Agency (ARIA) already identifies natural systems as an alternative and more sustainable form of computational structure to scale demand for compute. Programme director Suraj Bramhavar points out how a single session with ChatGPT uses 150x more power than a human brain conducting all its conventional functions. This discrepancy implies significant potential upside from engineering biological systems into electronic hardware solutions. Such a route also opens up possible synergies in the combined use of digital computing and the human brain.  

If the UK’s semiconductor industry is focused on design and instruction sets, the key asset would be the IP. As such, maximal government efforts would be necessary to safeguard this. A national cybersecurity team would be monitoring this constantly backed by the most powerful detection systems. On a timescale of decades into the future, such methods would need to be congruent with the latest quantum-proof/adjacent systems of encryption. This would have to be analogous to the physical security surrounding highly classified sites like Porton Down or Faslane which represent the most developed aspects of the UK’s deep state.  

As mentioned, the UK is poorly positioned to succeed in manufacturing conventional semiconductors in an era of generous subsidies from larger political actors. Yet there would still be opportunities in frontier aspects of this where no serious advantage has been built by anyone. One developing technology worth exploring would be in-space semiconductor fabs. This proposition sounds totally uneconomical, but this will likely become commercialised this century based on the two important characteristics. Firstly, operating in an environment of ‘zero gravity’ will increase yields. Without undue interference of gravity as felt on earth’s surface, materials will be able to be manipulated and purified with much greater ease. With reduced convection and sedimentation, crystals could be grown more predictably, and thus smoother and quicker. Chemical vapour deposition (CVD) and physical vapour deposition (PVD) processes will benefit from tighter particle aggregation. And assuming effective radiation shielding, space’s environment will reduce contamination compared with on Earth. The second opportunity of in-space semiconductor manufacturing is regarding autonomous self-replication. Software is already capable of reproducing itself, but this has not yet been achieved with full-stack hardware. Scaling engineering projects to planetary levels will require compact, modularised hardware that can reproduce itself totally self-sufficiently. That means a fully-integrated manufacturing of all components, with chips being the hardest part to create endogenously. By innovating in-space manufacturing to optimise chip production for earthly uses, an important by-product will be learning to simplify this process over the smallest dimensions, most basic inputs, and cheapest costs. It would also prove a useful opportunity to experiment with producing chips created by the UK’s new semiconductor design firms, especially designs which are optimised for self-replication.  

We can fairly assume that the economics of self-replicating hardware can become more viable in a matter of decades as a function of decreasing launch payload costs, advances in the cost and simplification of in-space semiconductor manufacturing, and through a positive learning rate that scales with the self-replication itself. If the UK can pioneer the research that cracks this, it offers it a head start in the development of Dyson Spheres and terraforming that unlock economic possibilities orders of magnitude beyond what humanity currently has. In the interests of space colonisation, terraforming would be particularly important on the statistical basis that there will be a large set of potentially hospitable planets (if subject to climate engineering), but there would be an extremely small number of planets, if not zero, which are already perfectly suited for human life. The UK’s success in this regard would depend on strong execution of its space strategy laid out above. It is also not clear how space law will continue to evolve as more actors and companies are involved outside of earth, especially in the context of self-replicating hardware that would operate mostly autonomously. By being the pioneering nation for in-space manufacturing (and LEO communications), the UK could establish itself as the regulatory hegemon going forward. It is crucial that the UK applies a common law framework early on firstly to continue with the current framework that has historically stimulated economic dynamism, and secondly as an opportunity to cement precedents that establish the UK as the incumbent rule-maker. I elaborate on this topic at the end of this essay, on how this system can scale as humanity moves beyond earth.  

These three state-backed ventures: conventional scaled production, alternative computational research, and in-space manufacturing, would all scale the UK’s ambitions in semiconductors. It is important they would be aggregated together into a British Semiconductor Group in the interests of mutually beneficial cooperation and ensuring a unified sense of policy direction. The BSG would be centred around Cambridge, where the majority of the country’s semiconductor firms are already currently based, transforming the city into a semiconductor free economic zone to facilitate this. That status would involve the area benefiting from full tax deductibility for IP, no duties on imported electronic hardware, and a unique regulatory set up that allows for highly experimental AI models and prioritised visa approvals for workers. Regarding salaries, it would be critical that bonuses awarded to R&D staff in BSG would be paid based on the performance of the wider group’s performance and not their department. This was a crucial policy at Apple that encouraged risk-taking innovations. If the individual manager is solely responsible for the profits and losses of his or her decisions in their vertical, in the face of risky bets they will naturally err on the side of caution which would be a failure of collective action to innovate aggressively. At no point should individual incentives regarding promotion or compensation conflict with the group’s aims to accelerate the rate of iteration and the boldness of the experiments.  

Under wider BSG strategy, Cambridge University would take on additional course capacity for areas relating to electronic and electrical engineering, material sciences, statistics, applied mathematics, and computer science. To absorb higher capacity, whole colleges would be created and solely dedicated to them. It should be an active effort of the UK to grow its engineering capacity under BSG by recruiting foreign students en masse. Under a scheme we can name British Advanced Semiconductor Programme (BASP), successful candidates who passed rigorous mathematical testing and security clearances from any corner of earth would be entitled to a fully-funded degree in one of the above areas, and with a generous stipend for living costs. After completion of their education, they would be obligated to then work for BASP, or other arms of the state’s R&D ambitions such as ARIA or Ascension-based aerospace engineering, for a period of five years in various fields in complex hardware or software based on their educational expertise. The country could reasonably expect to recruit 10,000 students per year to do this. At any one moment that would mean a state-backed legion manned with 50,000 foreign-born engineers, scientists, and mathematicians. Upon completion they would be awarded British citizenship and offered generous state funding for their startup ideas to incentivise them to permanently settle and develop their ideas in the country. With a sufficiently dense agglomeration of world-leading companies, UK startups, and extremely well-funded academia programmes, significant numbers would likely stay. 

The programme would need to be designed to prevent possible downsides. One would be the potential problem of other countries using the expensive programme as a form of free-riding to educate their own most talented youth at the UK’s expense before returning them to work on domestic projects. This would be overcome by having incentives for the candidates to stay in the UK as mentioned above, as well as clauses demanding the payback of the cost of education and research if there are violations of their contract. Issues could also arise from governments which are hostile to the UK sending individuals to act in their interests. This would involve rigorous SC and possible DV clearance procedures for candidates given their work on matters of national security or cutting-edge technologies. Additionally, candidates would be compartmentalised into certain specialist teams with minimal contact with other teams. This is common practice in large tech firms for good reason. Overall, the structure of the programme (culminating in an offer of British citizenship) would encourage candidates to become part of the country’s social fabric in a way that allays these two concerns. The US has an extremely strong record in its integration of foreign entrepreneurs. This is a culture that should very much be the aspiration of the BSG and UK as a whole. 

Downstream of the semiconductor industry, the UK has already established itself as Europe’s unchallenged centre for AI startups and worthy of global competition. The major bottlenecks remain weak capital markets and M&A regulation on the financial side, and lack of agglomeration in the industry as the best leave to the US. The exact details of financial reform is outside the scope of this essay, but that would be among the easier tasks of the UK government. The 50,000-strong BASP would be a sufficient draw for the world’s most in-demand talent and companies, based in an ecosystem based predominantly around Cambridge and London. At this point of economic development, the former would be merged into a mega-city for the UK’s capital. 

Successful pursuit of this semiconductor and AI strategy offers potentially limitless upside. Thorough coordination and execution across BSG could see the UK lead in semiconductor development, and establish the country’s dominance in the field globally. It offers scope for advances in tangential areas like quantum computing. As part of the broader AI strategy, the UK should also focus its efforts on data centre infrastructure design which Jensen Huang claims will be the new unit of compute. If this is synergised with chip optimisation, a positive feedback loop would develop as the UK built out the most advanced AI infrastructure, which in turn would generate the highest compute capability. As that broadens out from narrow intelligence to so-called ‘general intelligence’, such systems would in-turn contribute to their own development and push the boundaries of electrical engineering and all other R&D procedures across the UK. It is this unstoppable breakthrough that may prove the most important aspect of any execution of UK policy. Reaching this positive feedback loop could represent the so-called ‘[technological] singularity’ with the UK entrenched at its peak.  

Modular Nuclear Energy 

UK electricity consumption has declined about 20% since the turn of the millennium. With the shift from manufacturing to services largely having already taken place by 2000, and with the UK population having grown by 8 million people over this time period, this is symptomatic of a wider problem. Expensive energy has become a major bottleneck to higher economic output and living standards. This decline is out of the norm and unlike say, the US, which had record breaking electricity demand in 2022 (albeit their trend has been flattish since 2007). As of 2023, industrial electricity costs $0.41 per kWh in the UK, compared with $0.21 in France, $0.15 in America, and $0.09 in China. It simply does not pay to manufacture and produce in a country where it is so expensive. 

So what will enable this turn around? Fossil fuels are destructive and increasingly irrelevant on a long-term time horizon. The UK has neither the solar irradiance nor the manufacturing prowess to become world-beaters in photovoltaic technology – as extraordinary as that technology is. It does not have the geography for hydroelectric, especially regarding where population centres are based. Wind power had a troublesome 2023 and its learning rate of cost reduction has petered out in recent decades while solar efficiency roared ahead. These renewable technologies serve a useful role, especially solar, in improving the UK’s energy infrastructure over the medium term. Hydrogen fuel cell technology may become a major part of creating new energy abundance, although it faces poor prospects currently and is a bad bet judging by the current condition of the industry and forecasts; increasingly fewer names are predicted <$1/kg green hydrogen by 2050. Even in a bull case, a vast totally new energy infrastructure would have to be built from scratch if hydrogen is to power vehicles, homes, and factories. It is extremely combustible and its small molecular size makes it hard to store in easy and cheap ways. Assuming all of these factors can be addressed, it still leaves the grid requiring cheap and plentiful energy to power electrolysis. Ultimately, the worthwhile moonshot policy for powering the UK should fall to nuclear.  

The story of the UK’s prowess in nuclear science is a sad parable for the status of the wider country: a world-leader in the early and mid 20th century yet faded away by the 21st century. British contributions to nuclear science were paramount. Ernest Rutherford was a New Zealander by birth but in spent all of his higher education and adult life in the UK. His astonishing career saw him discover the nucleus (and with it a new model for the atom), identify alpha, beta and gamma radiation, and pioneer the transmutation of elements. He is deservedly buried in Westminster Abbey. In a similar time period, John Cockroft and Ernest Walton split the atomic nucleus in 1932 and later won the Nobel Prize for Physics. In 1941, the British (not the Americans!) initiated the first Allied project to build an atomic bomb. The British MAUD Committee initiated this project – the Tube Alloys programme – which later inspired and then merged into the US-lead Manhattan Project. The Brits went on to supply the Manhattan Project with crucial material and scientific expertise. British knowledge was built on the Frisch–Peierls memorandum of 1940 which had theoretically calculated necessary amounts of fissile materials to create nuclear weapons, as well as subsequent strategic and moral considerations. Head of the British delegation to the Manhattan Project was James Chadwick, the Nobel-winning physicist who had discovered the neutron in 1932 and also pioneered cancer treatment.  

In the civil realm, the UK created the world's first civil nuclear power plant, Calder Hall, which opened in Cumbria in 1956 at a total capacity of 200MWe. Energy generation was however initially intended as its secondary function; Calder Hall was primarily built to generate weapons-grade plutonium for the UK’s nuclear weapons programme. Since the nuclear heyday of the 1960s, much of the world has disregarded the abundant, clean, and safe prospect of atomic energy. The UK still remains 11th globally for nuclear energy output at 46TWh per year. France is third at 340TWh, and a reasonable initial target as a comparable country in economic terms. But unlike France, who built out their civil nuclear programme through a series of large regional power stations in the 1970s-1990s, the UK should pursue their competitive edge in small modular reactors.  

SMRs are a proposed design of fission reactors that can (mostly) be built in a factory and then sent for installation and use at a site. Unlike current nuclear power applications which are created and designed site-by-site as a form of infrastructure, SMRs hope to achieve a level of scale and standardisation whereby production economies of scale can be realised. This would help drive cheaper energy and safety costs on a per-watt basis, and result in negligible installation costs compared with building current nuclear power stations from scratch. Small reactors already exist in the form of ships and submarines with nuclear propulsion where there are no other alternatives. The important remaining question is whether the process of modularisation could make such technology competitive and useful as a baseload source of electrical power on land compared with alternatives. There are additional concerns over nuclear proliferation resulting from decentralised networks of more accessible fissile material. 

Rolls-Royce is the UK company at the forefront of SMR development. The engineering company speak of transforming ‘a large complex infrastructure programme into a factory built product’. RRSMRs will have a lifespan of 60 years that will produce 470MWe of electrical energy each in a three-loop pressurised water reactor (PWR). Ironically, the 470MWe output is considered too powerful to make it a literal ‘SMR’ under strict IEA definitions of sub-300MWe, but with approximately 90% of components being produced in factories it is still highly modularised. At somewhere between 20,000 – 40,000m^2 in size, it would be 5% the size of a conventional large-scale nuclear power plant despite producing 35% of their equivalent electrical output. Rolls-Royce are the main beneficiary of the UK’s Great British Nuclear programme launched in July 2023 to administer the competition and funding between different SMR designs. The six companies whose designs are being considered are Rolls-Royce, GE-Hitachi, EDF, Holtec, NuScale, and Westinghouse. In early 2024, new entrants joined the race in an MOU between X-energy and Cavendish Nuclear (owned by Babcock). Their Xe-100 design has not yet joined this official list being assessed by the regulators, but they have received £3.34m for designing 12 plants in the North East. 

Another key name, arguably the best in the world at its niche, is MOD-owned Sheffield Forgemasters. In February 2024 they successfully welded a SMR vessel demonstrator using local electron-beam welding (LEBW). It took just 24 hours to weld a structure in what would have normally taken one year. The company is on track to regain its ASME status to supply global civil nuclear markets according to strict regulations. In the meantime, it has positioned itself as the global leader in SMR welding technology having signed MOUs with X-energy, Cavendish Nuclear, Holtec, NuScale, Rolls-Royce, and GE Hitachi among others. Sheffield Forgemasters works adjacent to a wider government-backed research structure called the Nuclear Advanced Manufacturing Research Centre (NAMRC) which is innovating in several promising areas of manufacturing surrounding civil nuclear energy. Cavendish Nuclear’s AWESIM project which is partnered with NAMRC will reduce the time taken to assess welds from hours down to minutes thanks to acoustic, laser, and ultrasonic phased array sensors. Another new exciting innovation would be the use of CO2 as a supercritical fluid where it retains the flow properties of a gas while being sufficiently dense to act as an effective coolant. NAMRC are also experimenting with the integration of software and robotics to standardise machining and the production of fuel racks. This was associated with a ~40% reduction in times in early trials. Lastly, metlase allows for a more automated and uniform manufacturing process in the output of components for modular systems.  

London-based Newcleo is also offering SMR technology via lead-cooled fast reactors. There are two main designs, one at 30MWe aimed at large ships or small islands and the other at 200MWe as a form of baseload grid power. Both will be powered by MOX (Mixed Pu-U Oxides) consisting of depleted uranium – current nuclear ‘waste’ – and plutonium. Newcleo is also split across a network of funding, research and manufacturing centres in France and Italy as well as Britain. It is therefore revealing of the UK’s attractiveness when it is competing with similarly sized countries for different parts of the supply chain. September 2023 saw them snub Britain for France in building their initial reactor prototype. The UK’s GBN programme refused to back Newcleo’s AMRs on account of their alternative cooling and fuel systems compared with the RRSMR design. France also offered free land to build a prototype reactor, while the UK in its lethargic planning system still gave no decision two years after an initial request. This is an obviously solvable issue. While on the issue of design preference, the UK should be backing multiple SMR designs to maximise their chances of pioneering the optimal design whilst also stimulating competition that can accelerate the development timeline. 

This rigidity in preference for reactor design risks leaving the UK behind. Among relatively successful SMR designs so far include the work of Terrestrial Energy, whose SMR molten salt design operates at a higher temperature of 600 degrees Celsius increasing efficiency of electricity generation by 50%. It can also run off low enriched uranium, which bypasses the aforementioned issue of Russian control over the Haleu supply chain. Bill Gates-backed TerraPower claim they will be able to produce electricity at $50-$60 MWh through their sodium-cooled design running at 345MW that can operate at atmospheric pressure. Traditional designs of light water reactors have their drawbacks with lower thermal efficiency, high pressurisation levels, and inability to be fuelled by nuclear waste. Given the dozens of SMR designs being developed globally now, the UK cannot afford to fund all designs by themselves, but they can at the very least create a cooperative regulatory environment and competitive market to allow the winner to thrive. 

It is important to caveat British nuclear ambitions with the problematic condition of the uranium supply chain. In 2022, between one-third to one-half of all uranium ore used by the US and Europe was imported by either Russia or its ally Kazakhstan. The former controls about half of all global enrichment capacity, and almost all of its commercial high-assay low-enriched uranium (Haleu) capacity which would be necessary for many SMR designs, although not RRSMR’s. It is simply a matter of funding and time required to build up an independent nexus of enrichment facilities, coordinated together by the US, UK and key allies. Sourcing the ore itself is less of a problem given that Canada and Australia collectively supply 11,400 tonnes of the ore each year, which is equivalent to a quarter of global supply. It would also be worthwhile developing strong relations with fellow Commonwealth member Namibia, who are the third largest producer of uranium ore in the world. 

The OECD NEA’s 2020 report on cost reduction in civil nuclear programmes gives some key guidance for the future based on past nuclear policy mistakes and successes. Regulation should not change mid-construction, an approach that helped France keep their construction costs lower than America in the late 20th century. For a successful and rapid roll-out of SMRs that means having a regulatory environment fully agreed upon and prepared before SMR factories come online. That does not mean it can fully account for all eventualities or won’t need to be updated, but it does mean having a thoroughly researched backbone that can adapt to all issues as they arise be that accidents, safety, cost, export licensing, to name a few. This approach has the benefit of also establishing a global regulatory status quo determined by British authorities that can develop favourably to British SMR firms. In tech policy currently, the UK remains at the behest of the rulings of US and EU institutions and further weakens our standing in the sector (see: Democracy Technology Partnership Act, GDPR). The NEA report also quoted statistical evidence for a moderate learning rate that set French construction costs progressively lower over the course of separate phases of their nuclear programme. The CP1-2, P4-4' and N4 phases saw respective learning rates of 6%, 23%, and 19% in USD/kWe cost reductions over the total life of each phase. The modularity that is inherent to SMRs offers scope for an even stronger learning rate to develop over time as has most successfully happened with photovoltaic cells (Swanson’s law: a 20% drop for PV price for each doubling of volume) and lithium-ion batteries (also ~20%). Achieving learning rates on par with PV or Li-Ion batteries is almost certainly unfeasible due to the relative burdens of nuclear technology (regulation, bottlenecks in production, less scalable). We can however presume that with an aggressive expansion in SMRs, design and installation optimisations would lead to long term price reductions, especially in real-terms. This is a highly desirable alternative to the heavy infrastructure approach of British energy policy that sees a large project’s costs bloat uncontrollably at each new stage of assessment or expansion.  

A mature SMR market is still only something we can theorise. We can however assume the existence of a strong first mover advantage in operational terms and regarding regulation. A rapid, government-backed rollout also has the benefit of establishign dominance for UK firms early, making them hard to disrupt going forward and giving them substantial market leverage to fund further research. The SMR market will likely tend towards oligopoly (perhaps even monopoly) over the long run, as the capital needed to fund such large factories and expensive products could only be sustained by having substantial market share. The potential learning rate from scaling SMR solutions would be manifested by individual companies collecting empirical data, which could then be fed back real time to their development teams. Additionally, there would be 3+ sigma events and anecdotal evidence from end-users that could prompt further optimisations. The key premise is that being an early company allows for a head start on design optimisation that would drive costs lower. By growing market share early and quickly through this, a company should pass a point where their scale would be sufficient to fund their operations at a cash flow positive rate. The UK government should be aiming to ensure that in a mature SMR market all firms (or if a monopoly, the only firm) are domestic. That necessitates important requirements. Firstly, that the companies and the UK government aggressively protect real-time data to maintain the information advantage of their development programmes. Secondly, that benevolent oversight by the state, both in terms of proactive regulation and generous funding, not only allows but encourages rapid iteration of design that allows UK SMR firms to maintain and grow their competitive advantages.  

On a funding level it is a highly convex bet for the UK government to back SMRs. To make the UK the world-leader in the technology would require funding at levels around the single-digit billions per year. SMRs are indeed not fully proven as a technology. But that funding requirement is not cripplingly expensive, and the opportunity cost of not taking that risk involves a guarantee of the status quo: expensive energy, weak manufacturing and deep tech capabilities, and foregone economic output and exports. This funding should primarily be spent on subsidising R&D and production, buying up land in remote areas of the country for testing, and active but light regulatory oversight. The latter would involve regulatory officials working full time in cooperating with UK SMR manufacturers on-site rather than a Byzantine legal structure that is far slower and more restrictive for firms in an evolving space. This was a key success in Lockheed Martin’s R&D arm of ‘Skunk Works’, whereby they would request a project office be set up on the US government’s side as well as their own. The two respective project offices would be in dialogue all day, every day, which allowed Skunk Works to bounce ideas off their main customer to ensure constant and transparent communication. Such a method proved the gold standard for fast hardware engineering, and allowed for next generation aircraft to be built by teams 10x smaller than industry standards and in a mere one or two years. For UK SMR policy, conventional regulatory relationships with the private sector must be by-passed and a new UK government body must simultaneously fund, regulate, and empower SMR firms in-person with the ultimate incentive of stimulating rapid development and production.  

SMRs are indeed an unproven technology, but that is the risk one takes with pushing the frontiers of innovation. The UK cannot simply remain reliant on ‘proven’ technologies which involves others developing competitive advantages and scientific expertise ahead of us. Pushing a country to the frontier of hardware technology is an unavoidably risky and expensive endeavour but one that is worth it when the potential upside is extraordinary. Dominating the global market for cheap, near-limitless energy would radically improve living standards, technological capabilities, and the country’s power at a global level. A UK with cheap and abundant energy would be utterly transformed.  Being at the forefront of energy innovation gives an enormous advantage over countries that merely buy it after it has been developed. By being the first country to roll out successful SMR technology, the UK would gain the advantages of cheaper energy ahead of other countries. That would mean that for a period of at least a few years, it could be multiple times cheaper to manufacture in Britain over other similar countries. In the long run, costs would approximately converge as other nations obtain the same technology, ideally from British exports and licencing. Britain should still retain a noticeable cost advantage globally as it would charge licensing fees for its technology to other countries that would be reflected in their cost of electricity. Regionally, the UK could build fleets of SMRs across its eastern and southern coasts to export enormous amounts of cheap energy to the European continent. If the UK could maintain energy price competitiveness in the long run, it would attract widespread investment from energy-intensive heavy industry like steel, chemicals, and cement which (combined with high level R&D) would provide the necessary material for vast construction projects on earth and in outer space. As an example, Iceland’s competitive industrial electricity prices allows it to produce over 850,000 tonnes of aluminium each year. That is the 11th highest in the world out of any country, and over 90 times more than China produces on a per capita basis. For the UK, such cheap energy would also power compute at the national level and ensure there is no power bottleneck on training the most advanced AI models. The IEA predict that global power demand from all types of data centres will more than double to 1000 TWh by 2026. If Britain is able to pioneer SMR and its integration to data centre solutions, it would be able to capture a large share of that growth and subsequent compute capabilities. Modularised cheap power supply could suddenly make swaths of the country and its territorial waters feasible to mine for rare earths metals and critical supply chain elements. In making mass mining, production, and transport cheaper, SMR technology would be the UK’s opportunity to bridge humanity’s leap from being a Type I civilisation to a Type II one on the Kardashev scale.   

Galactic Common Law

In a scenario where the UK has mastered interstellar space travel, engineered dyson spheres through armies of self-replicating hardware systems, and built out vast infrastructure in nuclear energy and AI, a final question will be the one of governance. It is necessary to develop a legal system that can positively scale over such large numbers of planetary colonies and at such distances, while also maintaining a commitment to civil and commercial freedom. Historically, the UK’s common law legal system has been a pillar of the country’s development over the centuries. It empowered the mercantile classes in eras when most legal systems existed as a mechanism for entrenching absolutism. It promoted evolving, pro-social developments in the law that avoided the pressure build up that exploded into violent wars and revolutions elsewhere. The central premise of common law is that precedent cases established by court rulings help to guide future judgements under the principle of stare decisis. This does not mean judges blindly follow all previous cases, but instead build on them by adapting legal principles to new occurrences or even to deviate fully when justified. Principally, core constitutional tenets are developed dynamically and not statically. This incrementalism allows laws to evolve for social and technological developments unanticipated by lawmakers. Additionally, the law evolves over time with the contributions of juries which help to ensure that the legal system accurately reflects the attitudes of the population it seeks to govern, and to prevent centralisation of power among lawmakers and judges.   

Until the industrial and technological revolutions, large information asymmetries existed within states. It is incomprehensible to people now that before the internet, radio, or railway existed, information might take days or weeks to traverse a country. There is surely a causative element to the fact that the development of the printing press and the rise of absolutism coincided in the 16th and 17th centuries, and to the fact that the 18th century saw the further development of industrial technologies alongside the emergence of the nation-state. It is stating the obvious to claim that the internet has played a part in economic and cultural globalisation in the last few decades. As human civilisation has now achieved effectively zero-latency communication on our home planet thanks to manipulation of the electro-magnetic spectrum, the next step will be how that is managed as we start to move into outer space. For example, it takes 1.3 seconds for a signal on Earth to reach the moon, and vice versa. Any future human colony on Mars would need to wait between five to twenty minutes to receive messages from Earth. Humans have already reached the speed limit of possible communication; we just don’t realise this as a limiting factor when one side of the earth is only 1/14th of a lightsecond from its antipodean point. Yet a human based on one of Proxima Centauri’s planets (our nearest solar system) would have to wait nearly four and a half years until a message he or she sent would reach Earth. 

This situation will present logistical difficulties that mean that the largest scale of enforceable political centralisation will likely remain a single solar system where latencies can be counted in the minutes. Confederations could theoretically exist at the interstellar level but would be nothing more than a loose association. We maintain the theory of relativity’s assumption that information cannot travel faster than c, and we also assume that human travel will remain well below c. At scales greater than one light year, there is not much a civilisation can do to another one outside of its solar system, let alone know much about its ongoing events.   

Whilst this limit will consign day-to-day governance and flow of information to the level of a solar system, cooperation would still be mutually beneficial and possible at the interstellar level. When Britain sends colonies to foreign planets, systems should be established to maintain loose links at such long distances. Maintaining a common legal system could be a part of this.  We assume that there are infinite returns to scale for common law; that it can only be a net advantage to both increase the size of the body of precedent cases and to standardise it to the point of uniformity. This was realised as England evolved from a devolved, feudal medieval state into a centralised, industrial modern state. Implementation of case law was no longer subject to the whims of local judges and lords, but instead presented fairly in a more symmetric information market.   

There are several reasons as to why we ought to scale common law systems beyond the planetary level. Firstly, keeping common law systems at just the planetary level would forego the benefits of scaling large amounts of case data to the intra and then interstellar level. Secondly, the existence of a status quo common law system between British colonies across solar systems would help uphold a shared system of cultural norms in support of constitutional governance over large distances. That makes it a practical method to reduce the chance of any one rogue planetary colony subjecting our descendants to despotism. Thirdly, it is far more efficient for a colony to arrive on a planet with a pre-agreed set of pro-commerce legal traditions that allow it to prosper immediately, rather than to arrive on the planet without benefiting from the accumulated legal wisdom of their forebearers. Lastly, continuously sharing common law traditions between solar systems enables the legal system to be actively maintained and allows British planetary colonies that may have become tyrannical in isolation to be exposed to common law again and thus allow the possibility for the colony’s inhabitants to re-implement it in the event of a revolution. It is simply not sufficient to rely on single planet or intra-stellar civilisation to maintain common law in isolation and of its own volition. Future technologies will likely make it possible for malevolent authorities to totally erase entire areas of knowledge. Without the efforts of other planets mutually storing, developing, and sharing critical case law, then in the long run common law and its conferred benefits would eventually disappear. 

Wanting legal information to be repeatedly stored, developed, and shared invites the use of blockchain technology. There would be a central ledger containing all precedent cases among all British planetary colonies. A given case would be added to the ledger with information such as the events of the case, verdict, timestamp, anonymised jury information, and references to previous cases. Any time an update is made to the ledger by a planetary colony, it updates its own records before transmitting its updated ledger to all other planetary systems. There will not be any kind of coordination in the process of transmission or storage. Colonies in one solar system must not rely on the goodwill or political stability of another solar system to store or transmit ledger information on their behalf, for risk of corruption of key case law data (be that intentional or unintentional). Each colony would receive the constant flow of updated ledgers from other colonies (some of them decades old), and then aggregate them in a way that produces one centralised, corrected ledger that stores the maximum number of useful available cases. Corrupted or false data can be identified by comparing multiple ledgers and then disregarded accordingly. When iterated, this will allow for a continuously standardised and growing set of case law data that will optimise over time, and which would not be damaged in the long run even if some participants attempt to sabotage or abandon the process. It will also have the advantage of helping centralise what would become the ‘old’ English language, and facilitating a degree of cooperation and brotherhood when interstellar political unity would naturally deteriorate given the vast distances.  

Conclusion  

The relentless march of technological progress will continue, and with indifference to who may take advantage of its power. Nations that cannot advance their capabilities will remain at the mercy of those that do. There are myriad ways in which countries should better themselves, and so this essay’s policies cannot and should not be taken as an exhaustive list that solves all problems of the coming centuries. Though, the key proposals I set out are unavoidably interlinked. Better access to space will unlock cheap resources, energy, and semiconductors. Better semiconductors will power more advanced artificial intelligence to accelerate our hardware and software capabilities infinitely. Cheaper energy will strengthen these advances in AI, whilst unlocking cheaper mining and travel. Efficient legal systems will infinitely scale this under conditions of pro-commerce stability across all stretches of space and time. Future generations will look back at us in admiration of our strivings and in pity of our condition, much as we might look back at our forefathers in the same manner. Our descendants will orbit billions of planets and stars in far-flung parts of our universe, united by the common heritage of their language and legal systems. Harnessing the energy of the atom will power unimaginable abundance and prosperity, captured first from planetary sediment and then by harvesting fusion power directly from stars. Hyper-intelligent systems will drive endless progress in their economies and daily lives, powered by theories of computation not yet conceived.  This is the future the UK must claim for itself and humanity.