United States vs. China: Microchips, the secret weapon of a new technological Cold War

The technological dispute between the United States and China has become a geopolitical storm waiting to erupt. Both sides are investing, blocking each other, and moving their pieces to dominate the global chessboard. In the middle of this crossfire, many companies—such as chip manufacturers —are caught and are paying the price of this extreme rivalry.

The announcement of Xiaomi's first officially licensed premium mobile SoC , the XRing O1, is an attempt to reach the stature of giants like Qualcomm and Apple. And at the same time, it's a declaration of technological independence that shakes the foundations of the industry.

For this chip to become the brain of a cell phone, many stages must be overcome. As Lei Jun, co-founder and CEO of Xiaomi, points out, they have been researching the best connections for four years. With a team of more than 2,500 engineers, they have allocated approximately $1.67 billion and plan to invest $6 billion over the next 10 years.

To undermine Asian progress, the White House's trump card isn't a Sidewinder missile or a new tax, but a strategic software package called EDA . This set of tools, essential for designing next-generation chips, has become the most effective weapon.

By restricting the export of EDA software, Washington has taken aim at the heart of China's tech ecosystem. This tool is the digital equivalent of a master key . Without the right combination, companies like Huawei , SMIC, or DeepSeek face an invisible wall that hinders any aspirations of supremacy.

EDA allows millions of tiny components within a microprocessor to be orchestrated with surgical precision. Firms like Synopsys, Cadence, and Mentor—all based in the U.S.—dominate the sector, and because they are subject to U.S. law, they are key players in the geopolitical chess game .

So far, the details of the ban remain a mystery, although all indications are that the current licenses will remain in place. However, analysts warn that Chinese companies could be excluded from future technical support updates to produce in Taiwan using US technology.

Huawei, which seemed unstoppable in 2019, was the first victim. With a lightning-fast veto, the Trump administration blocked its access to key components such as advanced chips, electronic design software (EDA), and even Google's Android operating system. Now, this same pattern is beginning to repeat itself.

Xiaomi isn't the only company in the spotlight. Lenovo, which is also reportedly designing its own silicon, is starting to feel the pressure. Both are dangerously close to the sanctions zone where Huawei already resides, facing restrictions that not only stifle innovation but also disqualify it.

Without access to US-made instruments, Xiaomi is forced to replace TSMC as its manufacturer. The options are to rely on local suppliers like SMIC, develop its own design software, or find new substitutes.

The United States is not only trying to hinder the progress of “Made in China” chips, but also to restrict access to the most advanced semiconductors, which are no longer fractions of silicon, but the engine that drives AI and quantum computing .

Nvidia was in the eye of the storm: its graphics chips are the essential nutrient for fueling the most advanced generative AI. China, which accounts for 13% of its global sales, is a vital market, although still far from the dominance of the United States, which accounts for almost half. In this technological war, Nvidia doesn't just deliver processors: it sells power.

The real fear for the United States is that China doesn't even need the most powerful chips to make significant progress . The recent emergence of DeepSeek, a startup that developed an advanced AI model using limited hardware, set off alarm bells in Washington.

The message was clear: even with less power, Chinese ingenuity can make a difference. For the US, it's not just about limiting access to the best chips, but also about containing the talent and ambition that could make the most of them.

Demand for Nvidia's H20 chips is growing rapidly among Chinese giants such as Tencent, Alibaba, and ByteDance, TikTok's parent company. These chips, designed as a stripped-down version of the powerful H100 chips, became a sort of "plan B" for continuing to train AI models.

But Washington was quick to react: starting in 2022, it tightened controls, and in 2024, the Biden administration turned off the tap completely, blocking the sale of H20. In this silent war, even limited versions became too powerful to allow across borders.

This is the third time in less than three years that Nvidia has been targeted by these restrictions. Now, the California-based firm will need a license to sell its H₂O chip in China, a measure the U.S. Department of Commerce says is intended to safeguard national and economic security.

The company's CEO, Jensen Huang, did not sit idly by, calling the US policy a "failure," warning that the bans could push China to accelerate its technological self-sufficiency .

And indeed, while the US closes doors, China is redoubling its efforts to develop its own AI chips, investing billions and relying on giants like Huawei and state-backed startups.

Another facet of this prism is the measure seeking to regulate AI . The so-called "Magnificent Seven" went from demanding stricter rules to demanding complete freedom of operation.

The controversial "One Big Beautiful Bill" is a multifaceted proposal pushed by Donald Trump that, among other things, seeks to centralize control of AI regulation within the US federal government.

In theory, it seeks to ignite the innovation engine of American companies, providing a clear, level playing field for experimentation without government constraints. The goal is to maintain a lead over China in the race to dominate AI.

Even OpenAI, the company behind ChatGPT, not only endorsed Trump's AI Action Plan, but also used the occasion to launch a direct offensive against DeepSeek, the Chinese AI lab. In its presentation, it called on the government to ban these models, citing serious risks to privacy , national security, and intellectual property.

But if this law is passed, none of the "Union states" will be able to set their own AI regulations for at least ten years . This means that places that had already made progress in protecting their citizens from discriminatory algorithms, deepfakes, or privacy violations would lose their privileges.

The big winners in this mix are OpenAI, Google, Microsoft, and Meta . The national framework allows them to expand without regulatory friction. But while the giants celebrate, local startups and developers are relegated to the background, with less room to innovate, compete, or influence the rules of the game.

Instead of a diverse ecosystem, the risk is ending up with a digital oligopoly masquerading as progress.