Photo: Alexis Bienvenu © LFDE

By Alexis Bienvenu, Fund Manager, La Financière de l’Échiquier (LFDE)

Anthropic, the US AI company behind the Claude models, is preparing for an initial public offering. According to sources, market estimates value the company at between $1.5 trillion and $2 trillion, although the final range has not yet been set.

Moonshot AI, the Chinese company behind Kimi, one of the country’s most advanced models, is also preparing a possible IPO, according to the Wall Street Journal. Its latest known private valuation is around $50 billion, according to Bloomberg – thirty to forty times lower.

The gap in valuations is staggering, and all the more striking given that in January 2025, the performance of the Chinese model DeepSeek rattled investors, who feared the United States was losing its lead in the field. Barely two years on, markets no longer seem worried about an imminent shift in American leadership. So has China already lost the AI race?

On some fronts, the US advantage is indeed very clear. Today, Chinese models can compete with their American counterparts only on certain tasks. In programming, for example, the gap between the two sides of the Pacific is virtually non-existent: according to the LiveBench 2.0 benchmark of 25 October 2026, DeepSeek V4.1 Flash scores 96 points, almost level with Anthropic’s Claude Fable 5.1 at 97.7. But in long-horizon agentic tasks, the best US systems score close to 78 points, compared with 41 for the best Chinese model, in this case DeepSeek. That is a sizeable gap.

American companies have a second major advantage: their models are already embedded in a vast ecosystem of business software. An AI agent connected to Microsoft 365, Salesforce or Amazon Web Services, for instance, can, if authorised, access a company’s internal data in order to act on it. Chinese providers do not enjoy the same degree of integration with Western corporate systems, which are often reluctant to entrust them with sensitive data.

Finally, the United States has a commanding lead in computing power. According to Epoch AI, as of May 2025 the US accounted for around 75% of global capacity in the major publicly identified AI data centres, compared with 15% for China. It also enjoys privileged access to cutting-edge technology: Nvidia accelerators, TSMC chips, state-of-the-art ASML machines and advanced memory made by SK Hynix, Samsung and Micron. China, by contrast, faces restrictions that limit its ability to obtain this high-end equipment.

Even so, China is by no means condemned to a secondary role in AI. First, its models are significantly cheaper. According to LiveBench 2.0, a successfully completed task costs around $0.18 with DeepSeek, compared with several dollars for Anthropic’s most advanced models.

China also has one major advantage: the scale and speed at which its power system is growing. For years, the country has been investing heavily in renewables, coal and nuclear power, while the United States lags well behind in this kind of infrastructure. This momentum could support the expansion of AI, provided China can overcome the challenge of transmitting vast amounts of electricity between its main centres of production and consumption. In the central scenario of the International Energy Agency (IEA)[1], electricity consumption by Chinese data centres could rise by around 170% by 2030, compared with about 130% in the United States.

In short, America’s currently unrivalled dominance in AI is vulnerable precisely where China’s main strengths lie. In many respects, the two countries are mirror images of each other in their strengths and challenges.

The contest between the two powers is therefore far from settled. Judging by what China has achieved in industry over the past twenty years, it is entirely possible that a more frugal, less spectacular strategy – but one treated as a vital priority by a planning state – could prove competitive in the long run.

Disclaimer: These data and opinions, as well as the sectors and companies mentioned, are provided for information purposes only and do not constitute an offer to buy or sell securities, investment advice or financial analysis. The opinions expressed are those of the author and in no way engage the responsibility of LFDE. Past performance is not a reliable indicator of future results.

[1] Energy and AI report by the International Energy Agency (IEA), April 2025

Find all our Strategic Case articles