Main content

China - having its chips and eating them (without onions!)

Martin Banks Profile picture for user mbanks May 9, 2024
Is the use of sanctions by the USA that stop sales of state-of-the-art processor chips in turn being used by China as the launch pad for the next round of major changes in global IT?

china usa

A few weeks ago now the Financial Times carried a report which suggested that China is looking to play hard ball in the international world of integrated circuits.  The bottom line of the story is that China now seems to want only Chinese hardware and software to be used on its government PCs and servers.

It is an understandable position for any country to want to hold, though one that in practice seems to have little hope of achieving reality unless China sees its capabilities in these areas differently from what seems to be publicly known. It is also a story about which the average CIO or applications development team leader might be forgiven for thinking, 'OK...and?’.

Indeed, it does seem sensible to think 'so what?' and forget about it. Surely this is just China playing propagandist spoof on the stage of international trade for reasons best known to itself? How could this possibly affect how the average CIO or dev team leader in the average business goes about their daily life? The thing is, while this may not have an impact on the day job right now, there is a growing chance it could have a serious impact on a longer-term purview, and may well need to be planned for.

What China is saying

This may be just a part of the latest round of childish spats that politicians indulge in, but the Financial Times story reported on a Chinese document that listed the results of a ‘Safety and Reliability Evaluation’ study which seems to refer to those products that are considered safe for the Chinese government to use. The list includes chips, operating systems and databases, and that all of them should be made in China. This is linked to a national strategy for technological autarky, where the military, government and state sectors become self-sufficient. That means having the ability to produce, at scale, all the semiconductor processors and related devices, as well as all the operating systems and applications code, that they will be driving.

Key targets here are US chips, such as Intel and AMD processors, plus Windows OS and databases, such as Oracle, the globally ubiquitous tools for businesses and organizations of all sizes for many years now. The are severe doubts as to whether China could make suitable processors in volume, mainly because it does not have access to state-of-the-art production equipment, much of which is – or almost certainly will be soon – subject to the sanctions on such technologies being traded in China imposed by US President Joe Biden and covered last year by my colleague, Neil Raden.

Here, however, comes the big but - China has two obvious problems here. The first is its ability to manufacture the processors and related devices, which I will return to, the other is applications software. At the moment the applications from the leading providers of business management systems and services – Microsoft, Oracle and SAP in particular – are the mainstay of most businesses around the world. China is no different. A quick Google search showed those three held more than 60% of the Chinese market in 2019, so unravelling these from the state apparatus is bound to be both very expensive and time consuming. That, of course, is not necessarily an issue for the CCP (Chinese Communist Party). It plays by different rules to western governments.

Mature apps and AI

All of these applications are, to be polite, extremely mature. All have roots going back decades, and all are now huge 'onions', with layer upon layer of additional technology added to meet new external developments as they came along  - from mainframe (or PC in Microsoft’s case) onto client server, to cloud delivery, Tenanted Software as a Service and now containerized serverless operations. Such levels of complexity mean that, once established in the market, even updating them gets to be difficult, and changing to something different just too scary to contemplate.

For both vendors and users, therefore, the game has become to find other ways of adding new functionality and services, together with interfacing them with the established legacy. For China however that also means its stated objective of moving beyond these legacy systems to something different and Chinese is just so much hot air, unless it really does have an alternative on the grand scale/across the board/redo everything level.

Silly thought? Probably, and if that is the case then just about all of the rest of the Chinese story can be dismissed. For example, take the thought that sanctions should stop China getting hold of continued supplies of the latest chips. However, the country’s sphere of political and economic influence is now huge and there is certain to be a country somewhere that will welcome the business opportunity. It is not a new problem by any stretch of the imagination.

Consider, for example, the evidence in the 1970s about how Russia was obtaining chips from the USA. Back in those days, comparing estimates of the integrated circuit TAM (Total Addressable Market) for countries like Austria with their import statistics of chips made interesting reading. Then compare that with published import figures for state-of-the-art devices. This showed the latter were significantly higher. Where were they all going? Other searches showed up surprisingly high (for cold-war Europe) levels of trade between Austria and Czechoslovakia, Hungary and Yugoslavia, all then members of the Soviet Union.

China is therefore unlikely to go wanting when it comes to the devices it needs, not least because the country is known to have been stockpiling chips for two years or more. This is part of a wider geo-political strategy the country has been operating under for a good while now, where it seems clear the goal is to come out as the next global 'top dog’.

With that thought in mind, now ponder a simply stated, but globally significant thought - what if, as part of that plan, being the provider of the next wave of applications for business management, product development and design and so on would make a powerful, hard-to-resist complement to being global-political 'top dog'. And with the next economic growth surge likely to come from India and the African continent, where China is already 'best buddy of choice', being the provider of the machinery of those economies would be the application of `soft’ political power on a grand scale. Let’s face it, it has worked well for the USA for a good few decades.   

It is simple to say, 'Come up with an entirely new way of delivering business management’, but is asking a lot from any bunch of apps developers. Except that the arrival of AI, and generative AI in particular, is going to change the rules of applications development significantly, and is quite likely to bypass the huge 'onions' of legacy applications completely. Using AI to develop AI applications for business management is likely to create a whole new framework of business systems. They will use the same data, at least to start with, but will be free of the onion layers and able to head off in different directions that better suit the economics of the coming years.

And China is already pretty nifty at applying AI, so it is quite possible it is already a good way down the road to a new approach. Yes, it can face the world OK for the next five years or so with no direct support from the likes of Microsoft, Oracle and SAP, but would find itself increasingly hampered, and eventually hamstrung by the sanctions. It really does need a plan of its own making, so it is not beyond reason to think that one not only exists but is being implemented.

No more store-bought chips

Talk of applications being extra mature and long in the tooth can be equally applied to the processor chips that run the code. The gold standard here is the Intel x86 instruction set which lies at the heart of all the leading processors, regardless of manufacturer. Without that architectural commonality, all the leading applications would still be stuck in their parochial, proprietary and expensive siloes. It has meant that chips are now also increasingly large 'onions' with Intel continually tweaking the instruction set and adding to it to provide new facilities. Companies like AMD also tweak their version of the instruction set to add their own operational and performance nuances, but down at the heart, nothing changes.

AI has, for the first time, brought a significant change, with NVIDIA’s Graphics Processor Units being the processor of choice because of their speed at processing repetitive short instructions. But though demoted, the existing general purpose x86 architected processors are still needed for the admin jobs, so not much has changed. Intel is one vendor seriously working on hybrid chips that combine the functionality of both Graphics Processing and General Purpose devices, but it is possible to see them as just a stop gap till some new architecture is developed that really makes AI dance.

The chances must be high that this could come from China, it certainly has the intellectual smarts needed to make it happen. It might also be a new design that China can make in volume, not just for itself but also for its growing number of client countries. Indeed, some of those countries are certain to make ideal locations for additional production capacity.

Chip manufacturing is one area of known weakness for China. It can manufacture state-of-the-art devices, but only in low quantities, and the sanctions program means it is very unlikely to get the essential systems it will need from the Dutch manufacturer. This is, of course, one of the reasons China is keen to re-absorb Taiwan where the majority of global chip manufacture takes place.

But new AI-based applications architectures, coupled with new chip architectures and instruction sets could also point to the design of new devices that map more directly on to the technologies China has available and can readily work with. And it is highly likely that suitable substitutes for the Dutch production technologies, or reverse-engineered alternatives to them, will soon enough be appearing.

My take

The technological and developmental changes now charging at us are coming at such a pace that it is close to a certainty that the successful vendors of the future will not be the legacy leaders of the past and present. There is no reason why they should come from the countries that have led the way so far. With China’s expertise in chip design and manufacturing advancing fast, and the potential of its capabilities with AI both growing and becoming reality, the possibility of a change in global leadership is likely to mean that C-level executives in countries like the US and UK may have to make new friends fast, or be left behind.

A grey colored placeholder image