Tech Newsround – Asian exodus

China vs. USA – AI moves to Vancouver.

• The technology war is spreading with Microsoft reallocating employees from its lab in China to Canada and other South Asian countries competing hard to win the business that is leaving China.
• Microsoft established MRSA (Microsoft Research Asia) in Beijing in 1998 and it has grown over the subsequent years to become Microsoft’s 2^nd biggest research installation outside of the USA.
• Microsoft’s business is pretty small in China and with the Chinese government deciding to remove foreign software from its computers, this business is likely to shrink further.
• This beggars the question of why Microsoft would maintain such a large research facility in an economically unimportant territory, and I suspect the answer to this is geopolitics.
• China has been very keen (now more than ever) on developing its technological capability and having a facility like MSRA serves as an excellent training ground for personnel who can then spread their knowledge and experience more widely through the Chinese economy.
• From Microsoft’s perspective, I have always viewed this activity as a way to maintain a cordial relationship with the Chinese state, but now that this relationship is becoming increasingly strained, Microsoft seems to be reconsidering the value of this arrangement.
• To this end, Microsoft is moving a portion of the AI researchers that it has at MRSA to its new facility in Canada.
• Microsoft has stated that it is establishing a new lab in Vancouver that will be “operationally aligned with MRSA” and denies that there is any plan to deemphasise MSRA.
• However, actions speak louder than words and Microsoft has already shut down its LinkedIn business in China as it became much too difficult to operate with the state oversight that was required.
• Furthermore, Chinese companies are desperate to recruit AI researchers to help them create generative AI which combined with the other risks of operating on the mainland are making China less and less attractive as a place to conduct business.
• This is likely to incense the authorities as MSRA alumni make up an impressive list of the who’s who in Chinese AI.
• I suspect with the relationship between China and the USA at rock bottom, there is very little incentive to continue investing in China and so I suspect that this facility will begin to drift away from being a central part of AI development at Microsoft.
• Furthermore, other South Asian countries are working hard to attract business that is going to be relocating from China over the next 10 years or so.
• Vietnam and Malaysia have been particularly effective at persuading businesses to set up in their countries and Singapore is rapidly taking over from Hong Kong as the region’s financial hub.
• These are yet more signs of the decoupling trend that is already well underway and is going to continue for the foreseeable future.

Semiconductors – More cash please! Pt. II

• Intel has been at the forefront of demanding more subsidies from the USA and EU in order to build leading-edge fabs but this has hit a snag as the German government is unwilling to increase its contributions.
• Intel’s problems arise on two fronts which are that inflation is increasing the cost to build fabs more than expected and the fact that the fabs built outside of Taiwan, Korea or China are going to have a hard time competing economically.
• This is why Intel has gone back to the German government and increased its ask from €6.8bn to €10bn in order to build a leading-edge fab in Magdeburg.
• The problem right now is Germany is trying to find ways of spending less money rather than more and so this has split opinions within Germany’s coalition.
• Whatever happens, I suspect that the EU and the USA will end up having to offer more subsidies but even with these in place, it is unlikely that these factories will be able to fully compete with Taiwan.
• Hence, customers like Nvidia, Qualcomm and so on are going to be forced into a difficult choice as committing to leading-edge fabs elsewhere will be more expensive but will provide the insurance of supply being many thousands of miles outside of China’s sphere of influence.
• This latest stand-off is likely to delay the building of leading-edge fabs which in the current economic environment (excluding AI training chips) is probably a good idea.
• Hence, I suspect that slowing down the building of its new fabs probably also plays to Intel’s advantage given the outlook and its own difficult circumstances.
• I continue to think that the risk-reward trade-off of Intel vs TSMC or MediaTek or Qualcomm is sub-optimal and would need to see Intel a lot lower before considering getting involved.