"Europe is on the defensive when it comes to digital issues. What are the chances of getting out of the defensive in current tech developments from AI to online trading to mobility?
In Europe and also in Germany there are of course successful start-ups that are building new business areas, and many long-established companies are now quite advanced in digitizing their processes. In addition to the top dog SAP, we in Germany have a major player in Europe, at least in online trading, Zalando. However, apart from Spotify in Sweden, there are no market-shaping platforms that come from Europe. Anyone who has been reading this briefing for a while is painfully aware of this fact.
So the question is not, as some politicians might overzealously promise in speeches, how Germany will become the digital world market leader. Because we are structurally very far away from that. The question is how and whether the local business location manages to produce market-shaping companies. And with even more appropriate modesty: the question of how we can manage not to fall further behind internationally.
Let's look at three tech topic threads with this lens. One obvious one and two not so obvious ones.
Topic AI
Let's start with the dominating tech topic of "AI". Generative AI, and in particular but not only Large Language Models (LLM), has dominated not only the D:ECONOMY topics, but also the tech topics internationally this year. And rightly so.
The major relevant foundation models (GPT-4, Claude 2, Gemini) currently come from the United States. This is not immediately economically problematic per se, because these LLMs can all be licensed via an interface from the hyperscalers. OpenAI's models are available on Microsoft Azure, Gemini on Google Cloud, Claude 2 on Amazon Bedrock (AWS) and on Google Cloud.
The current technology also tends to be open to European companies. (Claude 2 is not yet officially available via API in Europe.) Nevertheless, the lack of large models from Europe is a potential structural challenge. LLMs are the basis for future technological developments and new products. European models could better reflect local values and cultural and economic characteristics, such as the many European languages. Fine-tuning of the models and, in general, the upcoming specialization of the large LLMs on specific tasks in the respective industries will depend on the ecosystems of developers and service providers surrounding the models. Such an ecosystem becomes more robust if the paths between the value creation levels are as small as possible. For example, if there is no ocean between the actors.
Two AI companies are currently making headlines in Europe. Aleph Alpha from Germany and Mistral AI from France. The Germans are developing a proprietary model and have raised a sensational 460 million euros for it. Mistral raised 385 million euros in 2023.
Is that much? Let's contrast this with a recent piece of news from the United States. According to Bloomberg sources, OpenAI is in the process of raising between $8 billion and $10 billion in new capital at a valuation of $100 billion. To date, OpenAI has raised a total of $13 billion in capital in the form of cash and cloud computing credits from Microsoft alone. Number two in the United States, Anthropic, has raised a total of $750 million to date. In addition, Google has pledged to invest another two billion dollars.
The European sums may make headlines here, but they are negligible compared to the American competition. This is important because training and operating the models is still very expensive. The training costs of LLMs are the largest, most obvious barrier to market entry on this topic.
The question arises here as to how the German flagship company plans to keep up with the same quality level as the American market leaders with a fraction of the resources. Data protection and data sovereignty as the only unique selling points will not be enough. The corresponding departments at Azure and AWS also implement this for the European customers of their services. It is a challenge.
While Aleph Alpha relies on a comparable proprietary approach to its American competition, the Paris-based company is taking the open source route at Mistral.
This is very relevant: open source can be a way to asymmetrically undermine American leadership. Europe can innovate rapidly here. Because in the United States the well-known battle between established companies (Big Tech, OpenAI) and innovators. This is being fought by the former, especially on the regulatory front. President Biden’s executive order on AI must be read against this background. This offers, at least theoretically, an opportunity for Europe to create an environment that is more innovation-friendly in global competition when it comes to AI. There are no local tech companies in Europe that stand in the way of this.
What is happening instead was recently summarized by the Financial Times with reference to the AI Act: "Like the Tyrannosaurus Rex in Jurassic Park, EU officials are chasing movement. If one part of the economy is growing quickly, they are quick to follow it. "
Regulation is important. But the timing seems bad for Europe. Instead of potentially suffocating open source providers, they should be strengthened. There is also the question of how sensible regulation can be made for something like LLMs, which even experts are still unsure how to categorize. Not to mention the challenge of anticipating future developments and their consequences. The EU is unnecessarily blocking a major opportunity for innovation here. Whether this will be final can be decided as early as 2024.
Trade topic
Digital platforms have long since reached all sectors of the economy. Some sectors deeper than others. A second current topic is trade, which is facing a major attack from China. The Chinese company PDD has achieved an international hit with the Temu discount marketplace. Temu is in first place in app charts worldwide according to analysis company Sensortower. It is not just Europe that has long since relocated the production of many consumer goods to China. Now the marketing and sales of the first affordable products are also coming from China. In addition to Temu, the fast fashion provider Shein is increasingly causing problems for the local fashion trade. (See our analysis on Shein and Temu.
Here, too, Europe could potentially become increasingly defensive. Especially when you consider that the marketplace architecture, which creates more scope for design in the rapidly growing marketplaces, is only embodied by Temu and other Chinese challengers, such as TikTok Shop. Please note our analysis of the two main strands of marketplace architectures.
The dynamics are still too easily underestimated here. Temu is a year old and already a household name in the US discount market, Shein has arrived at the center of European fashion retail and TikTok wants to bring online retail to its 1.5 billion users worldwide. 150 million of them are in the EU. This is not a side issue, even though it may feel like it.
While Europe may be retreating to the digital defensive position it has learned in this country even before the game has really begun, a potential shift in online retail has only just begun at the lower end of the market. A comparable dynamic is on the horizon for the heart of the German economy: the automotive industry.
Car theme
2024 is the year in which some Chinese car manufacturers such as the Geely Group will begin to bring their electric car models to European markets such as Germany. From a platform perspective, this is relevant because any successful form of digital mobility platforms will emerge where the installed base of hardware, in this case the cars on the roads, is highest. While the Dutch car company Stellantis' response to what it calls the "coming flood" is explicitly a low-price strategy, the Germans are currently increasingly relying on high profit margins at the upper end of the market, rather than on volume. Even Volkswagen relies on “Value over Volume”. This recently made German car manufacturers extremely profitable. But it also lays the first foundation for future market distributions in the car market that will significantly deviate from the current level, with all the associated consequences.
None of these issues will be finally decided in 2024. Of course, that’s not how economics works. But the decisions made in the coming year, whether in the economy or in the centers of political power, will be viewed in retrospect in a few years as very momentous." [1]
1.
Europa muss sich 2024 sehr anstrengen. Frankfurter Allgemeine Zeitung
(online) Frankfurter Allgemeine Zeitung GmbH. Dec 26, 2023. Von Marcel
Weiß
