#79 Deciphering China's Move Against Western Chipmaking Giants
Making Sense of “New” Chinese Chipmaking Breakthroughs, Should India target tariffs on digital trade?
Today, Satya S Sahu examines China’s recent procurement guidelines for public servers and PCs, which move away from Intel and AMD CPUs in the context of recently announced chip breakthroughs. Rijesh Panicker discusses India’s stance on applying tariffs on digital trade in the background of the recent extension of the WTO moratorium.
Also,
We are hiring! If you are passionate about working on emerging areas of contention at the intersection of technology and international relations, check out the Staff Research Analyst position with Takshashila’s High-Tech Geopolitics programme here.
Technomachy x Cyberpolitik: Making Sense of “New” Chinese Chipmaking Breakthroughs
— Satya S Sahu
In a “new” development in the long-standing US-China trade/tech war, China had reportedly issued guidelines for governmental procurement of laptops, PCs, and servers in December 2023, which will see a gradual replacement of US processors from Intel and AMD with homegrown alternatives.
A December 2023 document listing officially approved processors by China’s Information Security Evaluation Center was discovered last week by the Financial Times, which reported that these guidelines were “unveiled with little fanfare by the finance ministry and the Ministry of Industry and Information Technology.” The guidelines also mention a list of approved server and desktop operating systems and databases, which again do not include any Western options. FT reports that “government agencies and party organs above the township level” have begun following the new purchasing rules which include “safety and reliability” of processors and operating systems as primary criteria for procurement.
A detailed reportage from FT is available here. These purchase rules come in the wake of increased US scrutiny of tech and hardware exports of advanced chips and equipment to China, as well as in the background of the “3-5-2” and the “Made in China 2025” policies. The former is a 2019 directive that aimed to remove foreign hardware and software out of government and public offices at a rate of 30 percent by 2020, 50 percent by 2021, and the remaining 20 percent by 2022. That obviously has not happened. The latter was a national strategy presented in 2015, illustrating China’s ambitions to move away from producing low-tech, labour-intensive, cheap goods to more high-tech products and services, with a focus on reducing dependence on foreign suppliers as well. Semiconductors were an essential cog in this strategy, and is also the sector that has been hit hardest by US sanctions since 2019. But does Beijing’s approval of the 18 homegrown CPUs mean that Chinese CPUs are finally rivalling Intel and AMD’s offerings?
As Pranay writes,
Those bedazzled by China posed this as proof that China’s industrial policy is working. To them, these guidelines indicate that Chinese chipmaking capabilities are closing the gap with those in the West.
My takeaway was exactly the opposite. Because China has splurged massive subsidies to uphold an inefficient and not-so-effective local semiconductor industry, the only way to recoup that investment is to issue local sourcing mandates like this one. To me, these mandates indicate that local chipmakers aren't able to produce chips competitively. Because if they were, there would be no reason to issue such mandates.
Moreover, many such mandates have failed in the recent past. A so-called "3-5-2" policy in 2019 had domestic substitution targets to replace foreign PCs and software in government offices. At that time, the goal was domestic substitution at a pace of 30 per cent in 2020, 50 per cent in 2021 and 20 per cent in 2022. Back then, we saw similar reports claiming that China would close the gap with the West. But none of those targets were achieved.
Another reason for my scepticism is that had the CCP truly replaced Intel/AMD PCs; it wouldn't have quietly released this procurement directive; there would have been fanfare and chest-thumping.
One of the more interesting chip offerings listed on the procurement guidelines is by Zhaoxin, a joint venture between Via Technologies and the Shanghai Municipal Government. Their KX-7000 series x86 CPUs also coincidentally debuted in December 2023, featuring up to 8 cores and more importantly, utlising a chiplet design. Also coincidentally, leaked Geekbench scores for the KX-7000 were published last week, pegging the CPU’s performance somwhere in between Intel’s erstwhile Skylake and AMD’s Zen 1 offerings from the late 2010s. The CPU also apparently offers decent I/O with DDR5, PCIe 4.0 and USB 4 support. Combined with details of its purported performance, this means that for standard general purpose office work and governance services, these CPUs are more than capable options.
However, Zhaoxin has still not disclosed where the KX-7000’s chiplets are being fabricated and packaged, nor the production node that they will be using. Considering the fact that Intel and AMD’s comparable offerings utilised 16nm and 14nm nodes, Zhaoxin is also unlikely to be venturing into newer fabrication nodes. However, chiplet designs also mean that Zhaoxin needs to leverage advanced interconnect technology that allows the different chiplet dies (logic to memory or I/O) to communicate with each other, and that is a very difficult part of fabrication and packaging to get right. This merits a closer look as and when new details become available.
As of now, China’s Semiconductor Manufacturing International Co. (SMIC) is the likely candidate to fabricate these chips. Again, coincidentally, last week, Huawei and SMIC submitted patents for an etching method called self-aligned quadruple patterning (SAQP). This lithography method aims to produce advanced chips on sub-10nm nodes, obviously with a view to edge past US sanctions on chipmaking equipment that has limited China’s chipmaking abilities. Despite SMIC having reportedly made breakthroughs into 7nm territory last year, with Huawei’s Kirin 9000s mobile SoC held out to be the poster child for Chinese ingenuity and technological resilience in the face of sanctions, it is unclear whether yields from its production process have been competitive. SAQP technology was also initially tested by Intel in a bid to cross the sub-10nm milestone, but yield rates and performance/thermal/power metrics meant that it was ultimately abandoned in favour of more advanced processes using Extreme Ultraviolet lithography. This also lends credence to the theory that even if Huawei and SMIC manage to produce decent x86 CPUs, their cost per chip will render them commercially unfeasible in the consumer market, and therefore, we will see continued and possibly, more aggressive local sourcing mandates beyond just government and public sectors.
To judge the impact and extent of Chinese chip design breakthroughs, it’s a good idea to contextualise these announcements with the capabilities of the foundry supposedly making them. Longsoon’s 3A6000 is another example of a recently announced CPU. It is apparently competitive with Intel and AMD’s recent Raptor Cove and Zen 4 offerings, but it is based on SMIC’s 12nm process and will very likely run into similar issues mentioned above.
India is set to host the Quad Leaders' Summit in 2024. Subscribe to Takshashila's Quad Bulletin, a fortnightly newsletter that tracks the Quad's activities through the Indo-Pacific.
Your weekly dose of All Things China, with an upcoming particular focus on Chinese discourses on defence, foreign policy, tech, and India, awaits you in the Eye on China newsletter!
The Takshashila Geospatial Bulletin is a monthly dispatch of Geospatial insights for India’s strategic affairs. Subscribe now!
Matsyanyaaya x Cyberpolitik: Should India target tariffs on digital trade?
— Rijesh Panicker
At the World Trade Organization’s 13th ministerial conference in Abu Dhabi this March, India in a coalition with Indonesia and South Africa, strongly opposed the extension of a moratorium to outlaw tariffs in digital services. Despite a last-minute agreement to extend the moratorium, the expectations are that unless significant re-negotiations of what falls under the ambit of digital services happen, the moratorium will expire by 2026 at the next ministerial conference.
Trade in digital products was estimated at $8.2 Trillion in 2023, accounting for 54% of all services trade and growing at 8%. When the exemption was agreed upon in 1998, the entire digital goods and services area was in its infancy. No one then could have predicted the extent to which digital goods and services would dominate global trade. Indonesia has argued in a submission to the WTO in 2022 that emerging and least-developed countries have lost close to $56 billion in potential revenues from 2017 to 2020 due to the conversion of digitizable goods from physical to digital modes of transmission. In addition, there is also a concern that expansions to the original agreement have brought in goods that should not be under the ambit of digital services (e.g. MRI machines, which should not fall under the ambit of digital electronic transmissions).
A December 2023 paper from UNCTAD on “Digital Trade for Development” has argued that the customs duty losses from zero tariffs amount to no more than 0.1-0.33% of total government revenues. It also argues that these losses can be more than offset by domestic taxation on digital consumption in most cases. As per the paper, of nearly 105 regional trade agreements analysed, close to 100 have zero tariffs on digital trade, including agreements to continue with the zero tariffs bilaterally in the event of the moratorium expiring in the future.
Opposing arguments
The essence of the flip side arguments are as follows:
1. Developed nations benefit most from digital trade. Developing countries almost entirely bear the costs of no tariffs on digital services.
2. Developed nations are allowed to impose tariffs on certain sectors, such as agriculture and textiles, to protect their domestic industry, but developing countries cannot protect their fledgling domestic industries in the digital space.
3. As most of the economy moves to bits from atoms, we will need to learn how to tax them appropriately and imposing tariffs is a part of this. We should learn how best to do this now rather than extending an unclear and unfair moratorium.
4. The ITA expansion agreement has included services in the moratorium, resulting in developing countries losing flexibility offered under the GATT (including the ability to liberalise different sectors progressively)
5. Developing countries should have the ability to impose duties on specific digital goods and services to protect their own fledgling companies in these sectors.
Our view
Imposing tariffs on e-books and streaming services simply because they are end-consumption goods will only impost costs on Indian consumers and reduce choice in the long run. Also, where does this stop? Should we impose tariffs on online education courses offered by Coursera and MIT, too, because we want to let the IITs compete? In a world where knowledge is the key asset, we should not seek to impose costs on our citizens accessing such resources.
We must remember that the largest sources of digital trade are increasingly through a few intermediaries, which have become de facto platforms. Just as Indian consumers can access global products and services through these platforms, so can Indian companies serve global markets seamlessly. Imposing tariffs on these goods will go both ways and stop Indian companies from becoming world-class.
Given India’s focus on attracting investments in areas such as cloud computing, semiconductor design and manufacturing, our stance on this moratorium seems at odds with our goals. The fact that the moratorium might expire in 2026, will both deter multinational companies from investing in India and create incentives for other countries to plan their own tariffs in the space.
Even if we are able to push for import substitution in the digital products space by creating Indian alternatives, most of these domestic technology players will likely be part of globalised value chains. Ultimately, any revenue we raise by imposing tariffs on others will simply raise the cost of business for our own firms and expose them to retaliatory tariffs.
The logic of tariffs assumes that domestic players need preferential access to domestic markets to build their skills and compete better globally. However, as products are increasingly created natively digital, they will immediately compete and succeed or fail globally. Companies like Zoho and Freshworks show that Indian product companies can and do compete for global audiences. Tariffs will only expose Indian startups to duties and tariffs in other markets.
What We're Reading (or Listening to)
[Twitter Thread] Best Papers - Network for Advanced Study of Pakistan (NASP) Fellowship
[Video] When the Chips Are Down: A Deep Dive into a Global Crisis, with Pranay Kotasthane & Abhiram Manchi, by The Hague Program on International Cyber Security
[Podcast] The Seen and the Unseen Episode 374: Making Policy Fun with Khyati Pathak and Friends, by Amit Varma