Those who follow the tech markets closely know that 2025 was the year the narrative around Chinese technology changed dramatically. Until recently, the prevailing sentiment among Western investors oscillated between two extremes: on the one hand, the idea that China was destined to remain a perpetual laggard, on the other, the fear that Beijing was about to overtake the West on every front. As is often the case, the truth lies somewhere in the middle, but it's a complex truth that deserves in-depth analysis.

In this article, I want to share my reading of the Chinese tech situation, trying to separate the background noise from the real signals. I do so as a technology enthusiast who has closely followed the sector's evolution over the last twenty years, and as an investor trying to understand where opportunities are unfolding and where risks lurk.

The Turning Point: DeepSeek and the Efficiency Revolution

If I had to pinpoint a single event that redefined the perception of Chinese technology in 2025, it would be the launch of DeepSeek R1 in January. I still remember the morning I read the first technical reports on the model. As a computer scientist, what struck me wasn't so much the outright performance, but the underlying architecture and, above all, the declared training costs.

DeepSeek demonstrated that it is possible to train cutting-edge language models using fewer computational resources than American competitors. This achievement, which Liang Wenfeng, the forty-year-old founder and former financial analyst, made possible with the Mixture-of-Experts approach, sent Silicon Valley into a frenzy. Nvidia's collapse in January, with a record loss of market capitalization in a single day, was the most obvious symptom of the shock.

But the story didn't stop in January. Throughout the year, DeepSeek continued to release impressive updates. In September, V3.2-Exp arrived with DeepSeek Sparse Attention (DSA) technology, which halved operating costs. In December, the V3.2 model achieved performance comparable to OpenAI's GPT-5 on several benchmarks, while the V3.2-Special variant won gold medals at the International Mathematics and Computer Science Olympiad. The most significant thing for us investors? All of this is open source, released under the MIT license.

What fascinates me about this story is the paradigm shift it represents. The Americans built their leadership on privileged access to Nvidia's most advanced chips and the availability of seemingly unlimited capital. The Chinese, cut off from flagship GPUs by export restrictions, responded with architectural innovation. It's a bit like the story of the Japanese auto industry in the 1970s: forced by the oil crisis to rethink everything, they ended up inventing lean manufacturing.

Quantum Computing: Zuchongzhi 3.0 and the Qubit Race

While artificial intelligence (AI) is grabbing the headlines, there is another technological front where China is making significant progress, and one that could have enormous long-term implications: quantum computing.

In March 2025, researchers at the University of Science and Technology of China (USTC) unveiled Zuchongzhi 3.0, a superconducting quantum processor with 105 qubits and 182 couplers. The numbers they published in Physical Review Letters are impressive: processing speeds a million times faster than Google's most recent results, with world-leading fidelity metrics.

The interesting thing is that this hasn't remained a laboratory curiosity. In October, China Telecom Quantum Group and QuantumCTek commercialized a system based on the same architecture, connecting it to the Tianyan cloud platform, which already boasts over 37 million accesses from over 60 countries. Hefei, the city hosting these developments, is becoming a sort of Chinese "Quantum Valley," with dedicated industrial parks and dozens of specialized startups.

For long-term investors, this is a sector worth monitoring closely. We're not yet at the point where quantum computing generates significant revenues, but the roadmap outlined by Zuchongzhi's team envisions practical applications in quantum chemistry and drug discovery within the next five years. Whoever gets there first will have potentially insurmountable advantages in cryptography, materials discovery, and pharmaceutical development.

The Achilles' Heel: The Semiconductor Question

And here we come to the sore point, the one every serious analyst must address when discussing Chinese technology: advanced semiconductors. While China can boast real progress in AI and quantum computing, the situation is much more complex when it comes to manufacturing cutting-edge chips.

Let's start with the facts. The stated goal in the Made in China 2025 plan was to achieve 70% self-sufficiency in semiconductors by this year. The reality? We're around 30%, according to the most optimistic estimates. The 50% target for production equipment, while ambitious, masks a critical dependence in specific areas such as EUV lithography, ion implantation, and metrology.

SMIC, China's national foundry champion, is making remarkable progress. Using sophisticated multi-patterning techniques with DUV lithography (the kind they can still afford), they have succeeded in producing 7nm chips and are working on 5nm. Huawei, through HiSilicon, has developed the Ascend series of AI accelerators, with plans to produce 600,000 units of the 910C by 2025 and 1.6 million chips overall by 2026.

But there's a major catch. This DUV production is more expensive, with lower yields and technical limitations compared to the EUV processes used by TSMC and Samsung. It's sufficient for many current AI and 5G applications, but it doesn't allow them to compete on the technological frontier of the most advanced chips. US restrictions, which have blocked the sale of ASML's EUV machines to China, have created what Beijing calls a "technological iron curtain."

The crucial point for investors is to understand that this is a structural barrier, not just a capital issue. China is investing astronomical amounts in the sector: approximately $70 billion in the 2025-2026 semiconductor package, plus 400 billion yuan specifically for AI. But the expertise needed to develop EUV lithography from scratch requires decades of accumulated knowledge that money alone can't buy.

The Taiwan Variable: The Elephant in the Room

I can't talk about semiconductors without addressing the Taiwan issue. TSMC produces approximately 90% of the world's most advanced chips. It's a dependency that China views as an existential strategic vulnerability and that the West views as a key geopolitical lever.

I don't want to get into geopolitical speculation, but as investors, we must be aware that any scenario involving Taiwan would have dramatic effects on our investment thesis in the tech sector. A potential "reunification" would instantly resolve China's bottleneck in advanced semiconductors, but it would create global disruption on a scale hard to imagine.

My understanding is that China is pursuing a parallel strategy: on the one hand, developing domestic capabilities to gradually reduce dependence, and on the other, maintaining strategic options in Taiwan. For investors, this means that geopolitical risk must be consistently factored into sector valuations.

Implications for Investors: A Map to Navigate

For short-term investors, China's AI sector offers interesting opportunities, especially for companies benefiting from the open-source wave. DeepSeek's models are becoming a platform for others to build upon, and companies providing infrastructure for this ecosystem could see significant growth.

In the medium term, I would carefully monitor the "mature" chip sector. SMIC and other Chinese foundries are aggressively expanding capacity in nodes from 28nm and above, putting pressure on global prices. This is already impacting competitors like UMC, PSMC, and GlobalFoundries. Some are talking about an impending price war in the legacy segment.

For the long-term investor, the fundamental question is: will the technological bifurcation between the Western and Chinese ecosystems create two separate markets, or will a winner emerge? My hunch is that we will see a competitive coexistence, with China dominating in some specific segments and the West in others. Companies positioned to operate in both ecosystems could have a structural advantage.

Timeline: When to Expect What

Conclusions: The Paradox of Restrictions

There's an underlying irony in this story worth highlighting. American export restrictions, designed to slow down China, may have accelerated the development of an alternative technology ecosystem that is potentially more resilient in the long run. DeepSeek wouldn't exist in its current form without the need to work with less powerful chips. RISC-V wouldn't have the same traction without the urgency of finding alternatives to the ARM architecture licensed under Western law.

I'm not saying the restrictions were a strategic mistake, because the short-term impact was real and significant. But as often happens in the history of technology, pressure breeds innovation. And China's innovation over the past two years has been genuine, not just imitation.

For us investors, the message is clear: underestimating Chinese technology today would be as shortsighted as overestimating it was naive five years ago. The reality is nuanced, with real strengths (AI, quantum, legacy chips) and structural weaknesses (advanced semiconductors, equipment). Navigating this complexity requires granular analysis, sector by sector, company by company.

As always in markets, the best opportunities lie where sentiment is most polarized relative to fundamentals. And in Chinese technology, sentiment remains extremely polarized.