From the Space Race to the AI Race: Where Does Core Competitiveness Come From?

Humanity first landed on the Moon in 1969. In the half-century since, rockets, computing, materials, control systems and artificial intelligence have advanced dramatically. Yet returning people to the lunar surface has not become easy. It remains costly, slow and technically intricate, exposed to budgets, supply chains, political cycles and commercial risk.

This does not mean that today’s technology is inferior, or that humanity has lost the capacity for lunar flight. What has changed is the institutional setting in which a lunar programme must operate. Apollo and Artemis are not the same kind of undertaking. Apollo was an intensely political exercise in national mobilisation. Artemis is an attempt to establish a more durable arrangement in which public institutions and markets both have a role.

That difference does more than explain why returning to the Moon is difficult. It also offers a way to think about the present competition in artificial intelligence. When countries speak of foundation models, computing power, chips, open source and AI governance, they may appear to be comparing a product or a single technology. In reality, they are often testing whether a society can organise many different capabilities into a stable whole.

The Cold War space race is commonly described as a contest between capitalism and socialism. That is broadly true. But it is misleading if it suggests that Apollo was simply an achievement generated naturally by a market economy.

The United States was, of course, a capitalist country with powerful companies, research institutions and markets. But the Moon landing was not undertaken by companies pursuing an ordinary commercial return. It followed a national strategic decision, funded by the federal government, coordinated by NASA, and deeply connected to military technology, universities, the defence-industrial base and major contractors. When President Kennedy committed the United States to landing a person on the Moon within a decade, the point was not that lunar exploration already had a clear business case. It was to demonstrate American scientific, industrial, organisational and institutional capacity after the Soviet Union took the early lead in space.

The scale of Apollo makes this plain. Between 1960 and 1973, the United States spent about US$25.8 billion on Apollo alone—more than US$300 billion in current money. At its peak, Apollo consumed more than half of NASA’s budget. This cost analysis shows that it was not an ordinary research programme. It was a national allocation of resources at the highest priority.

In terms familiar in China, the United States was also, to a considerable degree, concentrating national resources to accomplish a major task. What it mobilised was not a single state-owned system, but the combined capacity of public budgets, the military-industrial system, private contractors, universities and research institutions. Companies participated, but they did not independently choose the objective of reaching the Moon. They carried out defined roles inside a framework of goals, standards, funding and contracts established by the state.

The Soviet case was more direct. Spaceflight was already part of its state planning and military-industrial system. The two countries had different institutional forms, but they shared one important feature in the lunar race: spaceflight was placed above ordinary commercial calculation. Its immediate purpose was not profit, but strategic prestige, technological leadership, military capacity and proof of institutional strength.

The real competition of the Cold War, then, was not about which system more closely approximated an abstract ideal of free markets or planning. It was about which system could, under intense external pressure and behind a clear national goal, hold together finance, industry, science, education and organisational capability as an effective whole.

Conditions are different now. With the end of the Soviet Union, the Cold War form of space race disappeared. The United States also finds it difficult to devote the same extraordinary share of public resources to one lunar objective for an extended period. Once Apollo was complete, the political urgency of the Moon declined sharply, along with the exceptional fiscal conditions that had sustained it. Humanity did not establish a continuous presence on the Moon not because rocket technology suddenly vanished, but because the mobilisation model of the 1960s was never easy to sustain indefinitely.

This helps explain why today’s return to the Moon needs a different model. Artemis is not trying to reproduce Apollo in full. It combines publicly led deep-space capabilities with commercial suppliers, international partners and the prospect of a future lunar economy. Government remains central: the SLS rocket, Orion spacecraft, deep-space communications, mission-safety standards, astronaut training and overall mission architecture still depend heavily on NASA and public funding.

Commercialisation does not mean that government has withdrawn, or that markets absorb all costs unaided. More accurately, government no longer seeks to develop and operate every system as an internal project. It turns some capabilities into procurement demand, inviting companies to compete, develop and operate them. NASA is working with SpaceX on the Starship Human Landing System and with Blue Origin on another crewed lander. It expects multiple providers to lower long-term costs, increase mission cadence and establish more sustainable lunar transport capabilities. NASA’s Human Landing Systems programme makes this objective explicit.

The Commercial Lunar Payload Services programme, or CLPS, illustrates the change even more clearly. Rather than always designing its own lunar landers, NASA buys delivery services from companies that take responsibility for the full chain—from launch and flight through to landing. The programme has a combined contract ceiling of US$2.6 billion through 2028 and involves multiple providers. NASA’s CLPS overview describes it as a way to enable faster, more frequent and relatively affordable access to the lunar surface.

But it would be a mistake to romanticise this as proof that commercialisation has solved the Moon-landing problem. Resource extraction, lunar manufacturing, long-term scientific bases, communications services and tourism all have potential, but most do not yet constitute mature markets. Many commercial space companies still depend on government orders. The state has not stepped outside the market. It is acting as the first long-term customer, risk sharer, rule maker and investor in foundational infrastructure.

Commercialisation does not remove risk either. It redistributes it. Fixed-price contracts transfer some cost and schedule risk to companies, but this can place excessive pressure on firms whose technology is not yet mature. NASA’s Inspector General has found widespread schedule delays in CLPS and noted that fixed-price arrangements created financial pressure for smaller and less experienced suppliers, including a provider that entered bankruptcy. The Inspector General’s review is a useful reminder that market mechanisms can improve efficiency, but they do not automatically eliminate the fragility of high-risk frontier engineering.

Artemis also has a more complex objective than Apollo. It is not simply to land people on the Moon, but to establish a more enduring presence near the lunar south pole, with a lunar-orbiting platform, landers, spacesuits, surface mobility, cargo systems and international arrangements. This is not a one-off sprint. It is a system intended to keep operating. Artemis II completed a crewed lunar flyby in April 2026. Artemis III is planned as a crewed low-Earth-orbit demonstration in 2027 to validate capabilities associated with commercial landing systems, with the next crewed lunar landing assigned to a subsequent mission. NASA’s current mission page reflects the reality of the process: complex systems advance through successive tests and accumulated reliability, not slogans.

Viewed from here, the AI question becomes clearer. The recent World Artificial Intelligence Conference in Shanghai, together with the World AI Cooperation Organization advanced there, has attracted signatures from 29 countries. The event demonstrates China’s wish to work with countries in the Global South through open models, technology cooperation and training. At the same time, major US AI companies had limited representative participation, while frontier models, advanced computing chips, cloud platforms and capital remain concentrated in a small number of countries and corporate systems. Reuters’ reporting on the conference and the organisation captures this tension.

It would be unwise to reduce this to a simple question of who attended and who did not, or to treat a conference list as a direct ranking of AI capability. China itself has significant companies and technical capacity, including Huawei and DeepSeek. What is worth observing is that AI cooperation, industrial ecosystems and critical technology supply chains are forming different networks. Some are built around advanced chips, closed models, cloud computing and capital markets. Others seek to build an ecosystem through open models, lower-cost deployment, application diffusion and technology cooperation.

The relevance of the lunar comparison is that AI competitiveness will not be determined simply by possession of one foundation model, one celebrated company or one impressive benchmark result. Foundation models are visible products, just as rockets and spacecraft are visible products of a lunar programme. What determines competitiveness is the underlying capacity to generate, maintain, improve and extend those products over time.

I am not a spaceflight specialist, but I have spent decades in IT, especially in software engineering and large software systems. My experience is that bringing an important product into existence—particularly one of genuine historical significance—is rarely just about delivering the product itself. The product is the visible result, but the work of making it often entails inventing a method, developing a new form of engineering organisation, sometimes even making a theoretical breakthrough and establishing a new operating model. The product is the concrete carrier of those deeper logics.

A country’s AI capability should therefore be understood as a larger system: basic research; chips and computing power; energy and data centres; software engineering; talent development; capital support; application settings; industrial coordination; institutional arrangements; and international rules and partnership networks. Any single element can be temporarily acquired, imported or subsidised. Long-term competitiveness depends on whether the whole can continue to operate, repair itself and upgrade under external constraints and internal pressure.

From the space race to the AI race, what is continuous is not the technology itself but a more basic question: can a society turn dispersed resources, knowledge and organisational capacity into the sustained ability to solve complex problems? The lesson of the lunar programme is not simply to pursue one spectacular success. It is to understand the system capable of producing further products behind that success. In that sense, we may be able to see more clearly what genuine core competitiveness is.