A Chinese startup launched an efficient artificial intelligence model that has sent tech giants’ stocks tumbling. What does this mean from a socialist perspective?
By Marcel Wando
DeepSeek, a Chinese artificial intelligence startup, recently launched an open source AI model that rivals the developments of U.S. technology giants and uses significantly fewer resources. This innovation triggered a historic stock market crash, with Nvidia, a leading supplier of AI chips, losing $589 billion in market value on a single day, January 27, 2025. To put this in perspective, this loss exceeds the gross domestic product of countries such as Portugal or New Zealand.
In addition to Nvidia, other U.S. tech giants such as Alphabet (Google’s parent company) and Microsoft also saw significant drops in their stock prices, reflecting investor concerns about the ability of U.S. companies to maintain their hegemony in AI in the face of Chinese advances.
These astronomical numbers are staggering, but what is really going on? In reality, it is nothing more than a speculative market correction in the face of a technological advance, which has exposed some of the contradictions of contemporary capitalism. But to understand this, we first need to know the facts.
The fluctuation of Nvidia stock
Nvidia is listed on the stock exchange under the ticker NVDA, with 24.68 billion shares in circulation. Each share was sold for $148 before the crash and now sells for $122. This fluctuation is the basis for calculating the 17% devaluation. However, 12 months earlier the value was $60.
The total mass of the nominal value of this capital rose from $1.48 trillion to $3.65 trillion and then fell to $3.01 trillion. That is, it rose by $2.17 trillion in 12 months and fell by $600 billion in one day. It is this last episode that everyone is discussing.
Since shares are securities that represent a fraction of a company’s capital stock, these figures would lead us to believe that the company’s capital has fluctuated by the same amount. That is not the case.
Speculation
The stock price is not necessarily the value of the company, but what investors think it will be worth in the future. For example, if a mining company exhausts the mineral reserves it has been exploring, its shares are worth nothing. At most, it is worth the sale price of its assets.
Now, let’s say that a year before the reserves are exhausted, the president of the company announces that he has found a new, purer reserve that will yield twice as much profit. Then he adds the caveat that the company will not be able to exploit it for another year. In this case, the stock immediately doubles in value because people are confident that the company will be worth twice as much next year.
Nvidia, in turn, is a hardware supplier for training and running AI models. In December 2024, the company had a 90% share of the market for GPUs, a fundamental product for the whole new Industry 4.0, but it is also active in the production of DPUs, CPUs, AI software and programming, and automotive chips. As there is speculation that this sector will grow significantly in the coming years, Nvidia was thought to be at the heart of this growth, becoming an almost absolute monopoly in the production of these goods.
Nvidia’s Real Capital
The difference between the stock price and the real size of a company is well known in financial markets. The earnings reports themselves do the math for us. This is the book value, the book of the company. It is the value of the company’s assets (real estate, machinery, stocks, cash, etc.) minus the value of its liabilities (loans, quotas, bonds, etc.).
The company has total assets of $96.01 billion and total liabilities of $30.11 billion, resulting in a book value of $65.9 billion. The book value per share is $2.69. This is 45 times lower than the share price after the drop on January 27 (TradingView).
Another way to estimate the real capital of the company is based on its annual revenue. In 2022, the company had revenues of $26.9 billion, while in 2023 they were $60.9 billion.
According to Greg Wu, a specialist in the semiconductor industry, the manufacturing time for chips varies depending on the technology used. In an interview with the Financial Times, he explained: “Typically, the whole process takes about 100 days on average. For 8-inch wafers, it takes between 10 and 15 weeks. For the most advanced technology nodes, it can take up to 120 or 150 days” (Financial Times).
This means that the capital installed in the factory allows between 2 and 3 production cycles per year. This means that for an annual income of 60 billion, the capital in motion is half or even a third of that figure. But let’s be generous and consider that the 60 billion is the value of real capital. It is also far from the value of the company after the loss on the January 27.
The BET of the stock market
Investors had high expectations for Nvidia, with revenues expected to grow by 265% in 2022 and 126% in 2023. For those who work, it’s as if their salary were to double or triple every year. This type of increase leads people to think: if this continues, where will I be in 10 years? On the basis of this speculation, the number of sales of shares in relation to their real value is determined. Before the January 27, it was estimated at 55 times, after that at 45 times.
These investors are nothing more than gamblers. Those who believe the company will grow buy shares. Those who believe the opposite do other things that make them money when the stock goes down (such as short selling). They even bet on what other speculators will speculate.
Therefore, the real value of the company does not change at all with this oscillation of the financial market. Production does not increase when stock prices rise, nor does it decrease when they fall. This is precisely because these prices are hyperinflated by financial speculation. Those who lose and those who gain from the fluctuations are the speculators. And, of course, the small investors are the ones who suffer the most because they don’t have the behind-the-scenes information that the big guys have.
The DeepSeek “bomb”
DeepSeek is a Chinese artificial intelligence (AI) company founded in November 2023. On January 20, it launched the DeepSeek-R1 model, which rivals advanced models such as OpenAI’s GPT-4, but at a significantly lower cost.
While companies like U.S.-based OpenAI spent about $100 million to train their models in 2023, DeepSeek claims to have trained DeepSeek-R1 for about $6 million, using about 2,000 Nvidia H800 GPUs, as opposed to the 16,000 GPUs typically used by Western competitors (The New York Times).
In addition to the cost of training, there is also a difference in the cost of operating this AI. When processing a given message, ChatGPT can cost about $0.0675 per 500-word response, while DeepSeek performs the same task for about $0.015, making it 4.5 times cheaper. It is estimated that operating costs are lower due to a strategy of selecting only the relevant parameters for each response, rather than all at once (Creolestudios).
In addition, the company hast kept its codes open, which means they are not an industry secret. This not only reduces mistrust about the veracity of its claims, but also forces competing companies to adopt its model. The significantly lower training costs allow for more competitors, which, combined with the lower operating costs, causes prices to fall. This creates a tendency towards capital depreciation for companies operating in this sector.
Frustrated expectations
Stock market investors expected Nvidia’s capital to be 55 times larger in the future than it is today. After the “bomb”, these expectations were somewhat frustrated, and now it is expected to be “only” 45 times larger. Stock market fluctuations like this are very common. In this case, it is only the magnitude of the fluctuation that has attracted attention, not the mechanism of the process. So this is not just a problem for Nvidia or the big tech companies, but a structural feature of financialized capitalism.
But this also shows that this is not a coup by the “Chinese communists” against U.S. imperialism. Nor is it an act of resistance by an “open source movement”. These are typical financial market movements, problems that are inherent to the capital market. It is an intrinsic effect of the anarchy of the capitalist market, it does not come from outside it.
The need for new capital
But in explaining the decline in stocks, we still need to explain why they were so high in the first place. Why are technology stocks priced so much higher than their real value?
It is normal for new sectors of the economy to be highly valued relative to traditional sectors that have already consolidated in the market. After all, the traditional sectors have already occupied their entire market, have fully developed their technological potential, and have a stabilized level of profitability. Only in the new sectors of the economy is there a new opportunity for growth.
Traditional sectors can even regain their value if they discover new markets, develop a disruptive technology, or create a new product. But these companies tend to grow not by expanding production or productivity, but by absorbing, merging or bankrupting their competitors.
Since the 2008 crisis, the world economy has been moving sideways or backwards. In the last decade, the growth of the traditional branches of capital has stagnated. More than ever, the bourgeoisie needs emerging capital companies like Nvidia.
The race for AI
In addition to Nvidia, there are dozens, if not hundreds, of other companies that are considered venture capital. They are in different fields, such as technology, finance, commerce, health, energy, mobility, infrastructure, education, media, etc. In general, they belong to the so-called Industry 4.0. The development of these technologies is the biggest hope the current imperialist bourgeoisie has to getting out of this recession. One of them is the Generative Artificial Intelligence.
The race for supremacy in artificial intelligence has intensified in recent years, with massive investments by corporations and governments. In the United States, the Trump administration recently announced the Stargate project, a public-private partnership that will invest up to $500 billion in AI infrastructure and includes companies such as OpenAI, SoftBank, and Oracle.
This massive investment not only drives research and technological development, but also creates a mystique around artificial intelligence. Presented as an inevitable solution to economic crises and an engine of unstoppable progress, AI is often treated as an autonomous phenomenon, detached from the relations of production and control exercised by big capital. This narrative obscures the fact that its advance is directly linked to the logic of profit and exploitation, further consolidating the dominance of large corporations over strategic sectors of the economy.
An uncertain future
At this point, everyone is wondering if the valuations of the future markets are still very wrong or if they will finally adjust to the value that will actually be realized. Will these companies really reach trillion dollar market capitalizations? Will it be born as a monopoly or will it be a broad market? Will the companies that dominate the sector be American? Today’s doubts are even greater than before the January 27, and the future is no longer the same.
Under capitalism, the future is always uncertain and chaotic. The lack of economic planning leads to the subjugation of humanity to the designs of an impersonal market. Optimistic investors are left with hope, pessimists with despair.
And what is there for a worker in this system? If they work, they support the profits of all these shareholders with their sweat, blood and tears. If they manage to invest a little, they either make mediocre profits when everything is going well, or they are robbed by the big investors when these crises occur. That is, unless they are unemployed or their country is at war, which is even worse.
Financial capitalism gambles with the future of humanity at the casino of speculation. Overcoming it is not a moral choice or a political preference, but a historical necessity to make a more predictable and prosperous future possible. Socialism is the only means we have at our disposal to put technology and the economy under the rational control of the working class as a whole, so that the economy serves humanity, instead of workers serving at the casino of speculators.
