China's "weird" Silence

THIS IS A STRICTLY OPINION POST.

China is one of those countries that makes people emotional. Some admire it, some fear it, some hate it, and many do all three at the same time. But my personal view stays simple: there are three things certain in life, death, tax, and China scoring Ws. Despite the constant suspicion, propaganda, mockery, and intentional downplaying it receives, China continues to progress, prowess, and impress. Yes, prowess is a word, and yes, I am using it. From manufacturing to infrastructure to robotics to AI, the speed at which China moves keeps forcing the world to pay attention. That is the backdrop for this post. Source

Rule Number 1: China always wins 🇨🇳

I genuinely believe in this quote: China always wins. Maybe not instantly, maybe not cleanly, maybe not in the flashy Silicon Valley way, but over time it keeps stacking results. Look at the recent run. China’s EAST fusion reactor, often called the “artificial sun,” set another major record in early 2025. Around the same period, images of what experts said appeared to be advanced new tailless Chinese military aircraft went viral, adding fuel to talk of a sixth-generation leap. On the robotics side, humanoids became a literal national spectacle, with Chinese robot firms turning the Spring Festival Gala into a flex of industrial ambition. This is exactly why I say China carries an aura. Every few weeks, there is some new moment that makes people stop and say, “Wait, they did what?” Artificial sun · Military aircraft · Humanoids

China’s catch-up on AI

This part is what fascinates me the most. America had a massive head start in AI. It had the top labs, the top chips, the top cloud stacks, the top funding machine, and the market hype all in its favor. And yet Chinese firms still managed to close the gap much faster than many expected. Stanford’s AI Index has noted that the performance gap between leading U.S. and Chinese models narrowed dramatically, and the 2026 report says the two sides have even traded places at the top of performance rankings at different points. Reuters also reported growing concern in Washington that Chinese open-source AI is creating a self-reinforcing advantage. So no, China may not dominate every single frontier of AI yet, but on value for money, open model distribution, aggressive iteration, and practical architecture choices, it has absolutely forced itself into the top tier. DeepSeek was the loudest example of that. Stanford AI Index · Reuters

China’s refusal to “bubble” AI

One reason I think the Chinese AI story feels different is because it does not look exactly like the American AI bubble machine. In the U.S., investors have often poured absurd sums into AI on the assumption that scale alone will eventually justify everything. In China, the environment looks harsher. That does not mean there is no state support, because there clearly is, but it does seem that many Chinese AI firms are being forced to show commercial traction faster, survive price wars faster, and justify themselves under tougher capital conditions. Reuters Breakingviews described firms like Zhipu and MiniMax as heavily loss-making and under pressure from squeezed venture funding, high R&D costs, and immature business models. So while the American side inflated AI with giant hype cycles and giant capex narratives, China’s firms seem more boxed into proving they can actually earn their place. That may be painful in the short term, but it could also produce harder companies. Reuters Breakingviews

China’s lagging behind

Now let me be fair. China is still behind the U.S. in two areas that matter the most: compute and chips. That is not anti-China talk. That is just reality. Estimates cited in recent reporting suggest the U.S. still holds a dominant share of global AI compute capacity, while China remains far behind. On chips, the gap is also real. U.S. officials and Reuters reporting have said Huawei’s AI chips still lag top American counterparts, even if that gap is narrowing. Huawei’s own CEO was quoted as saying its chips are one generation behind U.S. peers, though the company is trying to make up ground through systems-level optimization and cluster engineering. So yes, China has done impressive work in AI despite major bottlenecks, but the ceiling is still limited by access to top-end silicon and by raw compute scale. Models at the absolute bleeding edge still require chip and infrastructure depth that the U.S. ecosystem currently owns. Compute · Huawei chip gap

Something “weird” is happening

This is where the post gets interesting.

For a while, China felt like it had an aura-drop every month. A new robot moment. A new military reveal. A new AI upset. A new geopolitical move. But for the past few months, the vibe has felt quieter. Not dead, just quieter. And in AI especially, some signals look strange to me. MiniMax’s M2.7 release came out under a non-commercial license, which is not the kind of full-throttle open release many expected. At the same time, Zhipu has been raising pricing and is still posting very large losses despite fast revenue growth. Reuters reported that Zhipu raised API pricing by 83% while also disclosing widening losses, and Breakingviews painted a similar picture for the sector more broadly. I am not saying this proves Chinese AI firms are running out of money tomorrow. That would be too strong. But it does suggest pressure. Real pressure. Pressure to monetize, pressure to survive, pressure to justify the burn without relying on an endless blank-check loop. MiniMax license · Zhipu pricing and losses

The big question

So the real question is this: is China actually on the back foot, or is it simply working in silence on something much bigger?

This is where my instincts kick in. China is not a country that loves telegraphing everything. DeepSeek itself landed like a shock to global markets, with Reuters reporting that its rise helped trigger a brutal selloff and the largest single-day market-cap loss ever for Nvidia. That was not a small event. That was a reminder that one Chinese release can punch a hole through American certainty. And if you look at the broader pattern, Xi Jinping does not operate like Trump. Trump loves noise. China’s system often prefers concealment, pacing, ambiguity, and then surprise. That is why the recent quiet feels odd to me.

There are reports that China is expanding into unconventional infrastructure paths such as underwater data centres using seawater cooling. There are also persistent industry rumors around deeper domestic-chip optimization and future DeepSeek model development paths, including Huawei-linked possibilities, but those rumors are still not confirmed facts and should be treated carefully. What is confirmed is this: China knows chips and compute are the two big walls in front of it, and it is actively searching for ways around both. If it cracks even one of them at scale, the global AI market will feel it immediately. DeepSeek market shock · Underwater data centres

So what is it?

So what is it, then?

Is China struggling?

Or is China preparing something in the shadows, waiting to surprise the world again?

Based on my personal philosophy of Rule Number 1, I lean toward the latter. Not because China is magically unstoppable. Not because every rumor will turn out true. Not because every Chinese company is secretly winning. But because the pattern is too familiar now. People underestimate China, mock China, declare limits on China, and then China lands another statement. Maybe this quiet period is weakness. Maybe it is consolidation. Or maybe it is the silence before another shock.

We will see.