A Good Week for Chinese Innovation

Huawei has rolled out a new three-year indigenous AI strategy while China has finally cracked the Top 10 in the Global Innovation Index. America has failed to contain Chinese innovation.

This has been a very good week on the Chinese innovation front. China has not only said “no” to America’s Nvidia bullying tactics, but it also came up with a powerful AI counter response of its own. And it did so in the context of steady further increases in its overall innovation prowess as measured through the broader metrics of the Global Innovation Index.

All of this is consistent with several warnings I posted in July that can be found here and here and here. I am obviously dwelling on this point because it is central to the US-China conflict that continues to be the focus of my Substack platform. What started out as a trade war, has morphed into a succession of battles over new proprietary tech breakthroughs, technology transfer — with resolution of the TikTok skirmish apparently at hand — and an increasingly intense race for AI supremacy. In one critical sense, this aspect of the conflict is understandable: Innovation is where the rubber meets the road for the prosperity of any nation.

I continue to believe that the United States is on the losing end of its long-standing efforts to contain Chinese innovation. Huawei and its supply chain was the early target of so-called entity-list actions in Trump 1.0. In the Biden Administration, it was the “small yard, high fence” approach aimed at restricting Chinese access to advanced semiconductor chips. Now, in Trump 2.0, it’s all about the AI race, with on-again, off-again restrictions aimed at Nvidia’s most sophisticated AI processing chips.

It hasn’t exactly gone according to Washington’s script over the past several years. That’s especially the case when it comes to the recent battle over AI hardware. After initially restricting Chinese access to all Nvidia products, the US government buckled to company pressures and allowed selective Chinese purchases of less powerful Nvidia processing capacity. This prompted Commerce Secretary Howard Lutnick to gloat on CNBC, “We don't sell them our best stuff, not our second best, not oven our third best.” The strategy, if you want to call it that, is to get Chinese AI developers addicted to what Lutnick has arrogantly dubbed “the fourth one down.” Needless to say, such disrespect hardly went over well with the Chinese and their humiliation complex.

I have argued for some time (see Chapter 5 in Accidental Conflict) that America’s efforts at containing China’s technology gambit were bound to backfire, that Chinese companies and entrepreneurs, to say nothing of the government, would take American containment as a wake-up call that would evoke a strong competitive response. Indeed, that is how Huawei responded to the first round of tech sanctions in 2018 aimed at its market-leading smartphone business; within a year, it had rolled out a new Mate 30 smartphone in 2019 that was produced without any US components, and by the second quarter of this year had regained the top spot in China’s smartphone market.

Fast forward to 2025 and the same story appears to be playing out, as China aggressively responds to America’s AI containment campaign. First it was DeepSeek, China’s spectacular challenge to OpenAI’s market leading large-language search model, ChatGPT. Then, on September 17, Beijing announced a ban on all purchases of Nvidia’s AI chips, including the lower-grade RTX Pro 6000D that, as per Lutwick’s disdainful commentary, has been tailormade for China’s restricted market. According to press reports, the Nvidia ban was aimed at spurring Chinese indigenous AI chipmaking efforts.

And that is exactly what happened. Within 24 hours of the government’s ban on Nvidia, the still-sanctioned Huawei unveiled detailed plans for a three-year timeline aimed at developing a new "SuperPoD" computing cluster, using upcoming generations of its indigenous Ascend chips, that is aimed at competing with Nvidia’s high-end AI products. Notwithstanding predictable doubts in the West over Huawei’s ability to deliver on such an ambitious project, it pays to be mindful of Eva Dou’s conclusion in her recent opus, House of Huawei: “[Huawei’s] strength is its ability to accomplish titanic tasks at almost impossible speeds …”

The icing on a much bigger cake came in the form of the September 16 launch of the 2025 Global Innovation Index, a comprehensive ranking of about 130 countries that assesses innovation potential through the lens of some 78 key metrics, ranging from STEM-based education and R&D spending on the input side of the equation, to patents, trademarks, and scientific articles on the output side. For the first time in the GII’s eighteen-year history, China cracked the Top 10 in terms of the overall index, squeezing out Germany which fell to 11^th place (see chart below). That is an astonishing leap for a country that was ranked in 43^rd place as recently as 2010.

The output components of China’s GII ranking improved from 7^th to 5^th place, including the highest score of any country in the 130-nation sample for knowledge creation and diffusion. China’s score was lower for the innovation input components, but it rose from 23^rd place in 2024 to 19^th place in 2025, with especially high marks for infrastructure. By getting more bang in terms of innovation output from its relatively lower rankings on input, this is a distinct positive for China’s implicit innovation productivity.

In short, the US continues to delude itself in thinking it can contain Chinese innovation. China not only has brilliant scientists — many of whom are now returning home as America has been swept up in an anti-science, anti-academic, Sinophobic political wave — but it also has the full support of the State for basic research. Chinese entrepreneurs, who have been under intense regulatory scrutiny over the past several years, are once again focused on competitive breakthroughs in green technologies, electric vehicles, and batteries; if anything, the recent anti-involution campaign underscores the intensity of this new competition. And China has turned the full arsenal of its industrial policy apparatus to the support of advanced manufacturing and AI.

America’s response to Chinese innovation is counterproductive. It won't stop China, and even more importantly, it deflects attention and resources away from the heavy lifting that is required at home. Yes, the United States has plenty of leaders in the AI space, especially Nvidia. But as I wrote recently, I worry that the big LLM search companies, namely Alphabet, Meta, Microsoft, and Amazon, are collectively overspending on relatively low value-added AI data processing capacity. As we saw in the 1990s, this spending binge alters the cost structure of transactions-intensive services providers, transforming them unwittingly from variable- to increasingly fixed-cost producers. This is a classic set-up for a massive restructuring to eliminate the redundancies of excess processing capacity for the US AI industry as a whole. I also worry that the United States is in danger of squandering its long-standing lead in federal government support to basic research, while at the same time emasculating a powerful legacy of university support to scientific research.

Understandably, both China and the United States put equally high priorities on innovation. Yet there is no reason why both nations can’t make progress on the innovations front, without the other feeling as if this progress is the functional equivalent of an existential threat. China seems to be doing a better job of accepting the challenges that come with being second in the innovations race than the United States does in recognizing the inevitable threat to its leadership role. Swept up in the self-destructive interplay of Sinophobia and anti-science populism, America risks dropping the ball on innovation. By contrast, this week’s developments on the Chinese innovations front underscore the benefits that come from perseverance and strategic thinking.