Who wins the chip war?

In December 2024, US President Joe Biden issued new sanctions targeting companies exporting semiconductor technology to China, marking another escalation in the ongoing trade war between the two countries.

For years, the prevailing narrative has painted Chinese-made goods as synonymous with inferior quality or outright counterfeits of American products. However, this assumption is not entirely accurate. In fact, product quality often hinges more on the regulatory standards and market demands of the destination country than on the origin.

For instance, Chinese products sold in Egypt, often perceived as lower quality, adhere to different standards than the high-end versions sold in Europe, despite being manufactured in the same factories. Nonetheless, there are certain industries where China still lags behind its American and Western counterparts—one of the most notable being semiconductor manufacturing.​

In recent years, global efforts have focused on developing smaller, more efficient systems, with Taiwan and South Korea leading the race. However, supply chain crises and US-China trade tensions have shifted the competition towards another goal: domestic semiconductor production within their respective borders.

Unlike centralized industries, semiconductor production relies on a complex global network of components and technologies produced by multiple companies across various countries. For instance, the Dutch ASML dominates the production of critical equipment used in chip production, such as advanced photolithography machines.

Meanwhile, Taiwan’s TSMC and South Korea’s Samsung lead in semiconductor manufacturing and remain aligned with US export restrictions on advanced technologies to China. Additionally, the industry relies on raw materials sourced from multiple countries worldwide, highlighting the global nature of semiconductor supply chains.

This intricate interdependence and far-flung geographic spread make it all but impossible for any one nation to corner this critical industry. Sustaining production, let alone keeping pace with skyrocketing demand, requires deep global collaboration, bleeding-edge tech, and cross-disciplinary expertise.

Not all chips are equal

Microchips power everything from smartphones to cars, but not all chips serve the same purpose. For instance, the microchips used in automotive electrical circuits differ from those in smartphones. 

The current challenge goes beyond manufacturing more chips annually to address supply chain crises; it also involves producing the smallest and most efficient chips possible.

Smaller chips consume less power and perform better, but their lifespan is shorter. As a result, manufacturers of household appliances, for example, tend to use larger chips that perform reliably without the need for advanced, high-cost microchips used in smartphones, high-performance computing, and AI applications.

Currently, Samsung and TSMC lead the innovation race in nanoscale semiconductor fabrication, which is critical for developing next-generation GPUs—the backbone of advanced AI systems.

This field requires immense technological and financial resources, posing significant strategic challenges for Chinese companies aiming to carve out a space in this highly competitive field.

To join the competition, they must develop AI-driven solutions capable of competing with or surpassing the advanced models produced by American companies.

US tech giants face-off

Since the launch of ChatGPT, a fierce competition has kicked off among American tech companies to develop cutting-edge AI software. Elon Musk's xAI, for instance, has already invested billions of dollars in building a new data center, with more large-scale investments to follow. Similarly, Google has ramped up its investment in data centers to support its future AI systems.

This frenzy is nothing new to Silicon Valley, which has seen similar frenzies with every major technological breakthrough. From the rise of the internet in the 1990s, to digital platforms in the early 2000s, to the smartphone boom, and more recently, cryptocurrency.

Investors are pouring billions not necessarily to advance innovation itself, but to secure the largest possible share of a rapidly expanding market. Today, AI has become the latest battleground for this relentless competition.

Karma bites back

While American investors have poured billions into traditional AI development, a Chinese company has been quietly working on a more efficient model. This company recently unveiled DeepSeek, an AI model it claims to have developed for just $6 million—a fraction of the billions spent by US tech giants.

What sets this model apart is not just its low development cost, but also its significantly reduced operating expenses after training is complete. It is important to recognize that every query directed at models like ChatGPT or Gemini incurs real financial costs, primarily spent on electricity to power the massive servers running around the clock.

In contrast, the Chinese DeepSeek model demonstrates exceptional cost efficiency, incurring approximately one-thirtieth of the operational costs of its American counterparts.

Although this Chinese model is not yet capable of matching American AI models across all functionalities, it delivers comparable performance in essential tasks. This significantly reduces the need for vast amounts of semiconductor usage, a critical factor in the ongoing tech supremacy battle between China and the US.

Naturally, American tech companies will seek to adapt to this shift by either incorporating Chinese tech advances or developing similar innovations to cut costs. However, the immediate impact of this disruption is evident in global markets. Given the industry's volatility, smaller investors often absorb the heaviest losses, while larger investors are typically better equipped to weather financial setbacks.

Technological advancements are generally incremental, with companies introducing gradual improvements to existing products. For example, when Apple launches a new smartphone, Samsung follows up with a slightly enhanced competitor.

With DeepSeek, however, the leap appears to be far greater. This raises doubts about whether American tech stocks will recover in the near future, or will struggle to reclaim a significant portion of their recent losses.

A new form of bias

All AI models exhibit some degree of informational bias, a fact largely determined by the companies that develop them—whether users realize it or not. Notably, American models like ChatGPT impose stricter guidelines to enhance neutrality. For example, ChatGPT’s answers regarding the Palestinian issue have become more balanced and objective compared to several months ago.

On the other hand, Chinese models like DeepSeek operate under strict government-imposed restrictions, leading to distorted facts or the avoidance of sensitive topics. Try asking DeepSeek about the 1989 Tiananmen Square protests, and you will find it dodging the question with vague responses.

Therefore, cross-referencing historical sources remains the most reliable way to obtain accurate and unbiased information, especially on topics related to politics, history, or economics. Relying solely on AI can result in misleading outcomes due to the biases embedded in their design or training data.

That said, DeepSeek presents a compelling alternative by offering access to innovative technology at competitive prices, particularly for users facing financial constraints. Despite objections from US tech companies accusing China of data theft and intellectual property violations, the final choice remains with the end user.

As a user in Egypt, I see no real difference in who exploits my data—whether it’s the US or China. What truly matters is having an effective tool at the lowest cost, and in that regard, “Chinese is cheaper.”

---

(*)A version of this article first appeared in Arabic on Jan. 29, 2025