DeepSeek Aims for €64 Billion IPO on Star Market, While Washington Accuses It of Tech Pilfering
DeepSeek is preparing its IPO on the Shanghai Star Market with a valuation of €64 billion, tripling that of Mistral, while Washington accuses six Chinese companies of technological pilfering via large-scale distillation. This dual dynamic illustrates the growing fragmentation of the AI market between American and Chinese blocs, with major implications for regulation, employment, and access to technologies.
DeepSeek has mandated CITIC Securities to prepare its initial public offering on the Shanghai Star Market, with an estimated valuation of 64 billion euros — triple that of its French competitor Mistral. Simultaneously, Washington accuses six Chinese companies, including DeepSeek, Moonshot AI, and Alibaba, of “maliciously copying” AI model technologies developed in the United States, according to accusations relayed by US federal agencies.
Key points
A Chinese IPO against the global sentiment. While the climate around AI is deteriorating as Anthropic and OpenAI’s IPOs approach, DeepSeek chooses the Shanghai Star Market to raise funds for its computing infrastructure. The choice of CITIC Securities, a state-owned bank, signals a desire to secure financing within a domestic ecosystem, shielded from Western regulatory pressures.
The accusation of “large-scale distillation.” According to US authorities, Chinese firms are using the distillation technique to extract functionalities from American models. This accusation, brought by the FBI, NSA, and CISA, transforms a commercial dispute into a national security issue and could trigger new restrictions on the export of chips and software.
AI investments yield measurable results. Companies in the Valor 1000 ranking are seeing tangible gains in productivity, customer service, and decision-making, and plan to increase their budgets. This underlying trend contrasts with the growing market skepticism ahead of Anthropic and OpenAI’s IPOs.
Regulation diverges across blocs. An analysis of the “regulatory biographies” of GPT-4 and DeepSeek-R1 in the European Union, the United States, and China shows profoundly different regulatory access trajectories. Europe is moving towards a binding framework, Washington oscillates between national security and promoting innovation, while Beijing supports its champions while regulating uses.
The employment impact of data centers contested. In the United Kingdom, a think tank estimates that government and industry projections largely overestimate the creation of permanent jobs by AI data centers, due to automation. A debate that foreshadows social tensions related to the deployment of AI.
Context
The AI race has been structured since 2022 around a US-China axis, with Europe as a regulator and the rest of the world as an adoption ground. DeepSeek’s emergence as a leading player, with a valuation tripling that of Mistral, reshuffles the cards in a sector long dominated by Silicon Valley labs. The accusation of technological plundering is part of a long tradition of Sino-American disputes over intellectual property, from telecoms to semiconductors.
Key players
DeepSeek, a Hangzhou startup, aims to secure massive domestic funding via the Star Market. Washington, through its security agencies, seeks to protect its technological lead and justify new restrictions. Chinese giants Alibaba and Moonshot AI are targeted by the same accusations. In Europe, Mistral tries to hold its position against a competitor valued three times higher. Investors, meanwhile, weigh promises of productivity against fears of a bubble.
Data and figures
According to institutional data, DeepSeek’s valuation reaches 64 billion euros, compared to approximately 21 billion for Mistral. Furthermore, a 14-year study shows that a 10% increase in broadband adoption reduces waiting time by 1.19% for Americans, saving 1.5 hours per person per year, or 1.2 billion dollars annually. This correlation between digital infrastructure and productive time savings highlights the importance of investments in data centers, beyond direct jobs alone.
Analysis of stakes
In the short term, the American accusation could accelerate the fragmentation of the AI market into two distinct blocs, with separate supply chains. The targeted Chinese companies could face restrictions on access to advanced chips, slowing down their model training. In the medium term, the question of job creation by data centers will become a major political issue in advanced economies, as shown by the British debate. The winners will be the players capable of navigating between the two regulatory blocs; the losers, those too dependent on a single jurisdiction.
Prospective hypotheses
Scenario 1 — Regulatory escalation (probability 45%). Washington imposes new export restrictions targeting training chips and software. Indicator to monitor: announcements from the Department of Commerce and Beijing’s reaction.
Scenario 2 — Successful IPO and appeasement (probability 30%). DeepSeek raises significant funds, reassuring markets ahead of Anthropic and OpenAI’s IPOs. Indicator: Star Market performance and post-IPO valuation.
Scenario 3 — Sentiment correction (probability 25%). Anthropic and OpenAI’s IPOs disappoint, leading to a decline in valuations across the sector. Indicator: evolution of venture capital investments in the first quarter.
Why it’s important
The battle between DeepSeek and Washington is not just about tech giants: it determines access to AI tools for millions of businesses and developers, the cost of digital infrastructure, and ultimately, the distribution of productivity gains among nations. If fragmentation intensifies, European users could find themselves caught between two incompatible ecosystems. The question is no longer whether AI will transform the economy, but under what rules and jurisdiction.
This analysis was produced with the assistance of artificial intelligence, from institutional sources and verifiable open data. AI Transparency