The impact of DeepSeek on capital markets is far from waning even after the Spring FestivalIn the wake of this holiday, stocks related to the "DeepSeek concept" in the A-share market have surged by more than 46%. This increase has not only invigorated the sector directly associated with DeepSeek but has also positively affected related fields, including cloud computing, operating systems, and Huawei's HarmonyOS, as they all experienced a climb in stock prices.

In the Hong Kong market, short-selling pressures have eased significantly, with a noticeable uptrend across the boardThe Hang Seng Technology Index has been on the rise, helping the Hang Seng Index regain the critical threshold of 20,000 pointsAnalysts widely attribute this bullish sentiment in the Hong Kong market to breakthroughs in AI technology and a noted influx of investment.

While the secondary market responds visibly, there is an undercurrent of activity in the primary market, which seems to be brewing a storm of its ownSpeculation about the future of DeepSeek persists, undeterred by Alibaba's denial of any plans to invest $1 billion into the companyThis situation raises concerns about whether competitors like Kimi and Doubao can keep pace in this rapidly evolving landscape.

On a different front, companies like Nvidia, based in the U.S., are beginning to recover from the initial shockwaves triggered by DeepSeekThe narrative of a "collapse" regarding these firms is diminishing as the dust settlesHowever, the U.S. tech sector faces its own set of challenges, including strategic contractions, tariffs, and institutional layoffs, which also contribute significantly to the fluctuations in U.S. stock prices.

DeepSeek's meteoric rise is inseparable from a convergence of factors - the juxtaposition of domestic competitors, the missteps of adversaries in the field, and the dramatic declines in the stock prices of companies like NvidiaThese dynamic developments keep the market engaged and intrigued

Advertisements

A popular jest suggests that Huafang Quantitative, DeepSeek's parent company, could solve the perennial issue of AI profitability by shorting Nvidia stock before each upgrade of its AI models.

Despite the disruptive potential of DeepSeek, Nvidia's role as a dominant player remains unchallengedAlthough the costs associated with AI model development have decreased significantly, the trajectory toward lower computation prices is an inherent trend, as evidenced by the historical growth in other industries like the internet.

Even if DeepSeek hadn't emerged, other enterprises and AI models would still have overtaken that challenge, essentially lowering barriers to entry for computational powerNotably, Nvidia itself is a key enabler in driving down these costs, as indicated by its earlier efforts to optimize algorithm performance via its CUDA platform.

Market expectations regarding reduced computational costs are somewhat optimistic, believing that this trend will lead to greater overall demand for processing power - news that, for Nvidia, isn't necessarily detrimental.

Recent financial reports from major U.S. tech giants confirm an uptick in capital expenditures directed toward AI, with projections indicating that Google will invest $75 billion, Amazon $100 billion, and Meta $60-65 billion into AI initiatives this yearWhile DeepSeek may trigger cost savings and a slowdown in spending across the AI landscape, it is still a far cry from a catastrophic collapse.

Ultimately, the investments flowing into AI infrastructure will directly or indirectly amplify the demand for Nvidia's high-performance computing chips, which serves as a critical driver behind Nvidia's stock price recovery.

However, the emergence of DeepSeek has rendered the entire market more rational while the financial performance of America's tech giants remains a mixed bagWall Street estimates that the profit growth rate of the "Magnificent Seven" in the U.S. stock market could drop to about 22% year-over-year by Q4 2024, marking the slowest growth since 2023, and continuing a fourth consecutive quarter of growth deceleration.

The performance of tech giants has not met market expectations: Tesla's vehicle delivery volume and profit margins have declined, iPhone global revenue fell marginally by 0.9% year-over-year in Q4 2024, and growth for cloud businesses from Amazon, Microsoft, and Alphabet has also lagged behind projections

Advertisements

While current earnings can support the stock prices of these tech behemoths, it is notable that if these firms' dominance is seriously challenged, the possibility of a market correction looms large - financial institutions such as Goldman Sachs and Bank of America have already indicated heightened risks in the U.S. stock market.

The bull run in U.S. stocks is intricately linked to dollar hegemony and tech monopolies; concurrently, the stock market is seen as a crucial foundation for the national economy, meaning that tech firms from other countries inevitably face various forms of scrutinyAI, viewed as a powerful productivity tool that can be deeply integrated into multiple industries, has the potential for the U.S. to consolidate an absolute lead in this fieldShould it succeed in establishing a monopoly, the concept of an "AI tax," larger and more impactful than the current "Apple tax," is likely a future reality.

In response to the challenge posed by DeepSeek, the U.S. has strategically allowed its tech companies time to adapt, launch initiatives, and integrate DeepSeek into their operationsSo far, industry giants like Microsoft, Nvidia, Amazon, Intel, and AMD have all publicly declared their intention to harness the capabilities of DeepSeek's extensive model.

Amazon is integrating DeepSeek within its Bedrock and SageMaker services, while Microsoft is incorporating it into Azure AI Foundry, a move that requires users to pay for access to these platforms—essentially turning DeepSeek into a revenue contributor for these services.

Once U.S. tech firms fully understand DeepSeek's technology and are able to release models that offer superior costs and performance, their attitudes toward DeepSeek and similar Chinese AI technologies might shift significantlyFurthermore, any advantages the U.S. maintains in the high-end chip sector will likely prompt a tightening of sanctions aimed at restricting China’s access to these critical components

Advertisements

This trend has already started to manifest; on February 9, TSMC informed its customers in mainland China that all products related to the 16/14nm processes and below must be packaged at establishments authorized by the U.SDepartment of Commerce’s Bureau of Industry and Security (BIS), impacting all types of chips regardless of their applications.

Additionally, the U.S. government has requested clarification about the sale of Nvidia chips to DeepSeek, originally transacted through SingaporeThese actions clearly aim to impede Chinese enterprises from acquiring high-performance chips, ultimately stifling the progression of AI research and development in China.

An "AI NATO" appears to be quietly forming, deploying arguments revolving around data and national security concerns to pressure stakeholders into choosing sidesCountries like Italy, Australia, and South Korea have taken the lead in restricting DeepSeek's usage.

Liang Wenfeng, the founder of DeepSeek, recently received an invitation to speak at an artificial intelligence conference in Paris, exciting concern among domestic netizens regarding his personal safety, given the intense conflicts of interest that accompany these technological developmentsWhile no individual AI application can single-handedly dictate the world's complexities, Liang Wenfeng and DeepSeek have certainly begun to bear the heavy responsibilities of this moment.

This raises the critical question of whether to tie everyone's interests and fates togetherLiang Wenfeng, a remarkably private individual, seldom makes public appearances.

DeepSeek was birthed by a team of enthusiastic young technologists from the '80s and '90s with a relatively unadulterated background—remarkably, they are one of the few teams in the quantitative domain lacking overseas experienceHaving achieved financial independence, their initial motivations centered on pursuing technological breakthroughs rather than seeking short-term gains, generously opting to make their results open-source available to developers worldwide.

Yet, this scenario has shifted since DeepSeek began to gain traction

Now, consumers, developers, and investors are keen on reaping rewards from DeepSeek’s success.

Even more crucially, there is a shared desire for the platform to develop steadily without succumbing to the pressures of an "AI NATO" offensive or facing a collapseStability ensures that users can benefit from reliable Chinese AI services, instead of being met with constant notifications of "server busy, please try again later." The demand for functional systems remains high, leading many to see local deployment of DeepSeek as a viable, profitable venture.

Local deployment refers to the installation of DeepSeek's large AI models on personal computers, eliminating the need for internet or cloud services while allowing users to customize the model according to their specific needsBusinesses are particularly eager to pursue local deployment of DeepSeek, driven by their profit motives.

Since enterprises are deploying DeepSeek with their financial interests in mind, they naturally expect it to align with typical business logicIt is essential for DeepSeek to at least accommodate the operational and maintenance costs that arise from its wide application of AI models, ensuring the sustainability of its API services.

With the 45-day promotional period for the DeepSeek-V3 API service having ended, new pricing adjustments effective February 9 now set the rates at 0.5 CNY (cache hit) or 2 CNY (cache miss) per one million input tokens and 8 CNY per one million output tokensThese prices remain incredibly competitive compared to OpenAI's offerings.

Moreover, OpenAI recently secured a new financing round of $40 billion from Japan's SoftBank Group, with a pre-investment valuation hitting $260 billion.

In summary, capital remains a decisive factor influencing competition in the AI sector, whether it stems from user fees or equity financingAlthough Huafang is relatively affluent, it still pales in comparison to the scale of capital represented by entities like SoftBank.

The greater the scale of capital involvement, the larger the appetite for market acquisition

Advertisements

Advertisements