Olivier de Berranger

The birth of the Marvellous Three

This month’s Editorial by Olivier de Berranger, Chief Executive Officer and CIO, and Adrien Bommelaer, Fund Manager, La Financière de l’Echiquier

Sam Altman, CEO of OpenAI and creator of ChatGPT, has set himself the ambitious goal of increasing and, above all, securing the production capacity of the most powerful semiconductors. His plan is to secure sufficient production of the latest generation of semiconductors, essential for the development of generative artificial intelligence (AI). It is a staggering challenge, estimated at $7,000 billion by the head of OpenAI himself.

The semiconductor industry is at the heart of every major technological innovation. Without chips, digitalisation is impossible. Without chips, there would be no iPhone, no Facebook, no Tesla. Nor would we have modern defence. Chips are essential: they provide computing power and they provide storage capacity. With each cycle of innovation, a dominant ecosystem emerges. The PC revolution of the early 1980s was dominated by Intel, with the complicity of Microsoft. The smartphone market, 25 years later, was dominated by Qualcomm on ARM architecture, all manufactured in Taiwan by TSMC on ASML machines.

The current technological revolution, that of AI, will have a far greater impact on our lives than previous ones. AI is based on a new architecture, the GPU processor, which was first used for graphics cards and games, and then adopted more widely for new applications such as autonomous driving and machine learning. The success of generative AI, a more recent phenomenon, owes much to the intensive computational capabilities of artificial neural networks, which are executed more efficiently by the GPU architecture than by Intel CPUs.[1]

The dominant ecosystem of this new revolution has already been identified. Importantly, it is transnational. Nvidia in the USA designs the GPUs. TSMC in Taiwan is the only foundry with sufficiently advanced fabs[2] to produce semiconductors at the finest resolutions. Finally, ASML, based in Europe, is the only semiconductor equipment company capable of developing lithography machines, a key stage in the industrial process. The competitive advantages of these three players are such that they should remain dominant throughout this innovation cycle. From this point of view, they certainly deserve to be called the Marvellous Three! ASML, which has a monopoly on lithography, sells its most powerful machines for $250 million each, TSMC needs more than $10 billion for each new fab, and Nvidia has been developing GPU architecture for more than 30 years. We believe that earnings growth for these three companies will be strong and sustainable.

In our view, the potential risk to this ecosystem is not technological. It is elsewhere, highly geostrategic, as oil once was. Something is beginning to resonate: the nations that dominate computing power will have a major advantage. So, will governments’ protectionist reflexes undermine international technological cooperation that could benefit everyone? The Inflation Reduction Act in the United States provides $52 billion to encourage the relocation of fabs. Europe is launching its Chips Act and plans to spend $43 billion to build new fabs here.

China has long been trying to develop a semiconductor industry, with some success, but has come up against American efforts to block access to the most sophisticated ASML machines. Finally, if China were to take Taiwan by force, it would gain control of TSMC… but probably lose access to ASML. The idea of “happy globalisation” is a distant memory. The geostrategic situation is the main risk for this industry and in turn for the Marvellous Three!

 

Disclaimers: The opinions expressed in this document are those of the author. LFDE shall not be held liable for these opinions in any way. The securities and sectors mentioned above are given by way of example. There is no guarantee they will remain in the portfolio over time.
[1] Chips
[2] Plants where semiconductors are manufactured