Huawei’s Breakthrough: The Strategic Implications
Where do export controls go next? Is a credible Chinese AI chip coming down the pipeline? Plus: “Did Raimondo really endorse Huawei? And how much was her fee?”
On August 29, Huawei released its newest smartphone, the Mate 60 Pro, setting off a buzz in both China and the US. The device likely contains a 7-nanometer 5G chip manufactured by SMIC — a real achievement for Chinese advanced semiconductor manufacturing in the shadow of unprecedented US export controls.
Its release also just happened to coincide with Commerce Secretary Gina Raimondo’s visit to Beijing, making it a powerful geopolitical statement — Bill Bishop called it an “orchestrated snub to Raimondo.” I think he’s right. If Beijing didn’t approve of this release and was worried about the snub upsetting US-China relations, there is no way you would have seen the crowing endorsement on state and social media illustrated below. Beijing likes to talk about the importance of preserving “atmospherics,” and in the wake of the balloon reportedly convinced the Biden administration of the need to stop escalating restrictions on Huawei’s access to technology to put a floor under US-China relations.
On the other hand, had Biden pulled a similar stunt — like dropping the FBI’s balloon report during a Chinese official’s visit — Beijing would have portrayed it as unconscionable. Hopefully G7 policymakers take this incident to heart that for the China, “preserving atmospherics” is for thee and not for me.
Now onto the tech. First off, serious credit goes to Huawei engineers and the ecosystem the firm has helped prop up. We already knew that HiSilicon, Huawei’s design outfit, had serious chops: by the late 2010s pre-sanctions, they demonstrated the ability to go toe to toe with the big boys at Apple and Qualcomm. Now, after four years in the wilderness using what are outdated (or pilfered) ARM instruction set architecture and EDA tools, as well as last-generation manufacturing technology in DUV lithography, the firm produced something broadly competitive with other flagship phones.
What’s also impressive is how Huawei has broadly followed through on its commitment to indigenize production through aggressive investment in the Chinese semiconductor ecosystem. As the TechInsights teardown illustrates, nearly all the phone’s components come from Chinese firms. There is, however, one glaring exception: some rather sketchy SK Hynix memory (hopefully just) stockpiled in the run-up to sanctions. Recall that, after all, the US put China’s leading memory manufacturer YMTC on the entity list — and it seems as though the US has been even harder on YMTC than SMIC when it comes to licenses.
The outstanding questions below will determine just how big a deal the SMIC + Huawei breakthrough will turn out to be.
What’s the yield on this process, and will SMIC be able to produce 7nm at scale? As Doug O’Laughlin of Fabricated Knowledge colorfully explained to me, “‘Anyone’ could get a PhD to, atom by atom, build themselves a 2nm chip.” My guess is that the yield is not a complete joke: Huawei is a serious firm, and they would probably see more downside than upside in having rushed out a phone just for the memes, only to be able to sell a few thousand phones at an enormous loss. But it’s possible that this launch is not really ready for primetime, and instead was accelerated for the headlines to help convince the state, in its new $40 billion Big Fund, to not give up on pursuing leading-edge technology. But this strategy, as Ben Thompson points out, will prolong Chinese dependency on foreign tech:
The only way that China is going to break through the 5nm barrier is by going back not just a few years but a few decades: Shanghai Micro Electronics Equipment (SMEE), China’s answer to ASML, is building fully home-grown lithography machines using krypton fluoride lasers (Krf); this is previous generation DUV — first introduced around the turn of the century — using wavelengths of 248nm (current DUV processes use argon fluoride lasers (Arf) to achieve 193nm wavelengths). What SMEE specifically and China broadly needs to do is move down the learning curve for lithography. SMEE needs to learn how to use Arf, then immersion, and then EUV. It doesn’t need to take 20 years, either: while ASML needed to invent everything, SMEE has the luxury of already knowing what is possible; the same thing applies to every other part of the semiconductor value chain.
This, then, gets at why this PR cycle could be bad for China: the Mate 60 is a short-term propaganda victory that threatens to distract leaders who don’t understand semiconductors from facing up to and fully investing in the long road to semiconductor independence; it’s harder to accept that it might take 15 years when you have 7nm capability in hand right now.
Huawei and SMIC have a strong interest in convincing Chinese leadership to continue investing in the leading edge. Slowly developing market share with fully self-reliant low-end hardware doesn’t move the needle for Huawei, firm with nearly $100 billion in annual revenue. Rather, Huawei and SMIC aim to capture the sort of value available only to those selling frontier technology. Huawei in particular wants to gain back the market share it lost to Apple, Samsung, and Xiaomi in the wake of its entity listing — and it even aspires to compete with NVIDIA in offering accelerated compute.
Can pushing DUV to its limits unlock AI chips relevant in this technological paradigm? Designing and fabbing a frontier AI chip is a difference in degree, not in kind, from what Huawei and SMIC achieved with their latest chip. Absent tighter export controls, it’s likely that, as SemiAnalysis argues, “there are many scaled players and AI hardware startups that will soon be able to deliver on chips that are on par with Nvidia’s A100. These include Huawei, Biren, Tencent, Alibaba, Baidu, MetaX, and more. While we won’t dive into all their capabilities today, they will easily be able to deliver A100 class chips on SMIC 7nm in 2 years at significant volumes.”
Even without domestic fabbing, today, Chinese firms have access to NVIDIA’s H800 chip, which is good enough to garner billions in sales and enable them to buy access to bleeding-edge A100s via global cloud providers. That said, the amount of compute set to come online outside of China in the coming years is almost unfathomable — and the capabilities and cost of pseudo-self reliant chips in the coming years won’t compare to NVIDIA and Google’s chips. This in the near term it will make it difficult for Chinese firms to compete in a technological paradigm in which scale is essential to develop and deploy AI competitive with what the West has coming down the pipeline.
Will Huawei and SMIC’s advances ultimately prompt more aggressive export controls? The Biden administration chose to try to hold the line at 14nm. They worried that the Dutch, Germans, and Japanese wouldn’t stand for tighter controls, and during a chip shortage, tens of billions in Chinese capex to bring more lagging-edge supply on board seemed like a net positive, freeing up the CHIPS Act money to invest in bringing leading edge to the US. Further, policymakers probably assumed that DUV couldn’t be upgraded such that it could competitively make the chips at scale that could train frontier LLMs or weapons systems on AI supercomputers.
Many of these assumptions are now under question. US allies have proven more amenable to export controls than many assumed, allowing restrictions to expand into advanced DUV tools. China’s lagging-edge buildout — and the leverage that growing market share may afford Beijing in the coming years — is now a serious policy concern in DC. And SMIC is proving that it can do more with DUV than many had initially assumed. As EUV appears to be running out of steam, the gap between what can be done with the two lithography methods may close over time.
Chinese firms are making a risky bet building a leading-edge ecosystem on the back of Western tools from the likes of ASML, Applied, KLA, and Tokyo Electron. While this semicap is ostensibly sold into China today for use in 28nm and above, it’s likely being repurposed for leading-edge fabrication. But the US and its allies still have more than enough technological leverage over semiconductor chokepoints to stop China from taking the path of least resistance to producing leading-edge technology at scale. If the US decides to raise the thresholds on export controls or tighten licenses to prohibit servicing of tools already in China, what has already been sold into the PRC will brick out, leaving Chinese firms forced to re-engineer an almost certainly insurmountable swath of the semiconductor supply chain.
The question remains, will Biden (or a GOP president come 2025) ultimately decide to use this leverage?
The tradeoffs of a true technological blockade on the Chinese chip industry run as follows: More aggressive export controls are highly likely to dramatically set back the Chinese chip industry, perhaps irrevocably putting a decade or more between the global leading edge and China’s. PRC firms would have to take a major step back to recreate huge swaths of tooling and software they’re been able to piggyback on foreign firms for until now.
It would also once and for all make a lie of the deescalatory rhetoric the Biden administration pursued around “not trying to contain China” and pursuing a “small yard high fence” around tech (though it seems pretty clear that Beijing has already baked in this worldview and trying to convince this generation of Chinese leadership otherwise has well and truly failed).
The impact on global semiconductor firms could be considerable. While some demand would shift out of China, American firms would certainly face a hit to revenue and the dollars available to put towards R&D. That doesn’t mean, however, that in this new paradigm the gap between them and Chinese competitors would necessarily close. And the likes of Apple and Tesla would certainly face aggressive retaliation.
Pushing harder on export controls may involve moderately upsetting allies, though documents like Germany’s new China strategy speak to an openness to a more ambitious tack. “Defectors” will try to de-Americanize their technology (like apparently what happened with the A510 ARM core Huawei used which was exclusively developed out of Cambridge and Sophia-Antipolis) to access the Chinese market with uncertain success. Technological progress more broadly would also probably slow down. For instance, there’s no way we’re getting Vision Pros without the sort of US-China collaboration which would come to a halt in this new paradigm.
Last year, I argued that with the October 7th restrictions the US has crossed a point of no return on US-China tech competition. Pursuing, as Sullivan characterized it, “as large of a lead as possible” on critical technologies like AI and semiconductors may require more aggressive decoupling as this chip illustrates how Chinese firms are able to close the gap in the current framework where the US only tries to limit their access to the highest end tech.
Ultimately, which is the worst risk profile for the US to adopt—if calibrates its export controls too lose or too tight?
When SMIC first fabbed 7nm chips, the firm tried to keep it low-key — but reporting on this breakthrough in the summer of 2022 helped push through the October 7 export controls and the CHIPS Act. Such a headline-grabbing announcement from Huawei could backfire: politicians like Rep. Gallagher are already using the announcement to pressure the administration to tighten controls. Dropping this phone during Raimondo’s visit was a high-stakes move on the part of Huawei and the CCP. It surely provided a shot in the arm to Chinese stocks and a boost of techno-national pride in a moment of serious economic questions. Even so, if the Chinese tech ecosystem isn’t prepared for the tighter export controls that Mate 60’s release may engender, the phone’s release may prove to be a cheap high not worth the crash.
I’d encourage you to check out the latest SemiAnalysis piece I collaborated on which goes deeper into the capabilities of the chip and potential responses the Biden administration could consider to alter the trajectory of the Chinese semiconductor industry.
A look into Chinese media’s coverage of Mate 60’s release reveals that chipmaking prowess is now deeply entangled with narratives of national strength and geopolitical competition. To be sure, the jury is still out on how sustainable or competitive China’s chip manufacturing will be in the future, or what this all means for the long-term effectiveness of US semiconductor policy. And concerns about the future of supply chains continue to linger. Even so, the Mate 60 Pro is being celebrated as a triumph of Chinese resilience in the face of American power.
Below are translated excerpts from leading Chinese analysts discussing what Huawei’s breakthrough means for chipmaking in China. They cover:
Comments from Chinese semiconductor insiders on how Huawei pulled off this coup, including reporting around yield and design architecture in the face of US sanctions,
The rise of national security in popular Chinese tech discourse;
How Huawei’s “anti-marketing” campaign succeeded;
Some excellent Raimondo memes.