Private valuations for China’s DeepSeek have risen nearly fourfold since the beginning of the year, underscoring the escalating price war associated with generative AI platforms. The latest estimate for the Chinese large language model (LLM) coincides with ChatGPT developer OpenAI commanding a $850 billion valuation, up from $500 billion at the end of last year, while Claude’s developer Anthropic is now marketed at $900 billion, compared with $350 billion over the same period.

Yet, despite following similar repricing trajectories and comparable technical proficiency, DeepSeek’s estimated valuation of $50 billion represents only a fraction of Anthropic’s or OpenAI’s figure. Analysts interpret the valuation gap as indicative of diverging paths taken between US and Chinese developers where the former have embraced a capital-intensive spending strategy ladened with economic implications, while the latter are focused on application deployment. 

Given AI’s near-limitless possibilities, few expect the current spending cycle to taper off soon. For the year ahead, US tech conglomerates are expected to dole out $700 billion for AI-related infrastructure, representing twice the amount spent in 2025 which itself was double what was allocated in 2024. Aggressive expenditure for data centres and semiconductors has already emerged in the economic data, with Jefferies estimating that nearly 1.15 percentage points of the US’s first quarter GDP growth of 2.7 percent, or 43 percent of the expansion, can be attributed to AI capex.

[See more: CUHK launches Hong Kong’s first full-stack embodied AI laboratory]

Demand for Chinese made semiconductors and other advanced components has even spilled over into trade figures, with Morgan Stanley estimating that half of China’s 14 percent export growth in April can be attributed to overseas AI spending. This phenomenon was also visible in Hong Kong’s first quarter GDP reading, where the 5.9 percent jump was partly driven by the 23.8 percent gain in exports, highlighting the broadening depth of the AI windfall across the region. 

Conversely, non-AI spending remains subdued, raising concerns that a pullback in the current AI-related capex wave could quickly usher in an economic downturn. Analysts view this risk as being more manageable in China, where AI-related expenditure, like this year’s $85 billion dollars, adds just 0.2 to 0.3 percentage points to annual real GDP growth.

A deployment focus 

While the global AI race is often framed around model supremacy, analysts view China’s emphasis on deployment as its next structural economic pillar. By leveraging open-source, lower-cost models like DeepSeek, smaller enterprises face fewer adoption barriers, noted Hersh Oberoi, director of the global research department at Balfour Capital, in comments sent to The Bay. 

“Cheaper AI broadens access, and broader access is what ultimately drives economic impact,” he says, a comment backed by Morgan Stanley’s assessment that a meaningful AI uptake could lift China’s total factor productivity growth by 3 percentage points over the next decade, which would add 3.5 percentage points to its potential GDP growth by 2035. 

Despite early-stage commercialisation hurdles confronting AI developers, for Chinese companies deploying cost efficient LLMs the surge in AI inference is already enhancing operational metrics and triggering higher market multiples. Morgan Stanley estimates that earnings before interest and taxes (EBIT) margins for AI integrated companies have risen 13 percentage points since 2021 to 17 percent. This in turn has also lifted forward earning projections by 60 percent, demonstrating that investors are increasingly migrating to businesses integrating AI adoption strategies as Beijing is expanding capital channels to homegrown tech firms. 

Market investors have responded. Moore Threads, a Beijing-based chip maker seen as a direct competition to Nvidia, quadrupled its market value on the first day of trading and is now worth $50 billion while Knowledge Atlas Technology, developer of Zhi Pu, an open-source LLM similar to DeepSeek, has risen almost ten times since going public and is now worth $70 billion. 

[See more: Hong Kong graduate job vacancies fall 61 percent as AI takes over]

From the sidelines, skyrocketing prices are increasingly difficult to dismiss, Oberoi remarks, noting that some like OpenAI are being valued at revenue multiples beyond the traditional software sector even as they project continued operational cash burn over the next few years. 

“However, unlike the dot-com era, today’s leading AI companies are generating real revenues and driving measurable enterprise adoption, rather than relying primarily on investor enthusiasm and momentum,” he explains. 

Continuing deal flows bode well for private companies like DeepSeek as China’s entire AI supply chain of software developers and chip manufacturers becomes domestically and vertically integrated. Even if China’s inference capabilities rely on suboptimal semiconductors, analysts argue that because those upfront investments were lower, their potential returns on investment for Chinese AI remain attractive.

DeepSeek’s reported $50 billion valuation follows the platform’s expansion into multimodal capabilities, equipping its flagship chatbot to handle not just text but also images and video. The company became a household name in early 2025 after the release of its lower-cost R1 platform rattled global stock markets as it cast doubt over the sustainability of the AI arms race. However, the more recent launch of its V4 model lacked the same effect, reflecting the competition the company faces from domestic rivals like ByteDance’s Doubao, which already offers multimodal features, reinforcing how the competitive landscape for AI models ultimately benefits its users.