Ep 2: Oct 24 – European crisis and AI boost. Is it the perfect storm or also an opportunity?

September just ended, we saw many signals in the European market regarding crisis, and all over the world, all the challenging situations are causing markets to experience high fluctuations.

The German automotive company VW, for the first time in 75 years, has come to the idea of closing two production sites in Germany. At the same time, Intel, again in Germany, planned to stop (formally hold for 2 years) the investment to build a new factory of chipsets (for 30B$) due to its current crisis, putting in trouble the plan for Europe to reduce by 2030 its dependency on outside EU borders production of chips.

There are clear indications that the electric cars market competition from China for European automotive and the acceleration of AI and SoC chips from competitors of Intel have been driving such a situation. At the same time, as a consequence of the overproduction of semiconductors after COVID demand and for electric cars, as this demand slowed down and AI chipset demands rose, several chipset producers started to focus on the AI shift, in some cases, quite reorganizing, like STMicroelectronics, in some other cases, like Nvidia, just accelerating an already existing focus. In the last few years, we saw a shift to own production of chips for many US tech enterprises, especially with the proliferation of SoCs. However, I like to focus on some correlations of technological events here to look more at the forward trends rather than focusing on the root cause of the past.

During September, Mario Draghi, as consultant to Ursula Von Der Leyen, delivered his one-year analysis report about European future competitiveness, indicating also key weaknesses and areas to invest. He spoke about a 750B-800B$ per year spending expected between 2025 and 2030 to remediate to a situation that, if not properly addressed, will lead to the progressive agony of the European community by 2050 at the latest. This approach included a difficult-to-digest and implement unique market strategy and faced resistance due to the overall cost of the approach. However, his report is useful to identify some key highlights to take away. 

There have been defined three key aspects to focus on going forward for Europe’s evolution: innovation, decarbonization, and security.

It’s interesting to see, in the detailed report, that the innovation is seen not only as important by itself but also as influencing the decarbonization. At the same time, security is highly linked to increasing the production within the EU borders to reduce dependency on other countries, which was indeed in the focus of the EU, for example, for the chipset production in Europe by 2030, which will most probably not be achieved.

Statistics say that the EU reached the peak of population, and predictions are that by 2050 will be a decrease from 742M to 703M or 5% in population. Here you can refer to the same Draghi’s report or to, for example, Population Pyramids.

Digging into Draghi’s analysis is quite clear that the cloud providers’ competition is lost for Europe. I also share this perspective, considering the multiple years of investment that brought those hyper-scalers to reach the actual maturity. In his report can be found that cloud providers in Europe represent just 2% of the cloud market, where over 65% is mainly in US from the three major hyper-scalers (Microsoft, Amazon, Google). Where he sees synergies is around the AI development in specific areas that, being in an early stage of maturity, cross the world, could give the chance to Europe to step up its presence, even if depending on computer power from the big hyper-scalers, all US-owned. 

In the report, there is the suggestion to use the power of HPC from Europe (mostly from big research centers) to use it for AI calculation, but I doubt this solution will bring the same value once will start to compete with other services, including data integrations, typically part of the integrated and modular solutions coming from the big US hyper-scalers. I feel that dependency on the US hyperscalers will not go away due to the variety and elasticity of components available in such environments.

An interesting aspect is about Digital Transformation in AI time:

• In this report, Draghi also projected a 60-110B$ revenue increase in the market of Pharma from the optimization coming from the drugs analysis augmentation, thanks to AI as one of the key aspects. Also, McKinsey reported the same trend in January, giving a clear idea that in AI we have a good real opportunity to improve for such types of businesses more mature in advanced data elaboration.
• Clearly, in the AI development, it will be key to focus on the P&L impact of the AI and its cost, as also pointed out by RussellReynolds. Some areas, like robotic process automation with AI, are giving some improvement from automation and reduction of redundant activities, and are achievable for many enterprises, but have limited value increase. The difference starts with work augmenting, typically from creating process improvement, including a human digital twin, where the augmenting will start to ramp up the value creation for the enterprise, but will require a bigger effort to be rolled out in all those businesses still in an early stage of maturity. In this area, we see interesting movements coming also from some historical ITSM vendors extending their capabilities toward a platform approach to build integration toward business key data and analytics, and use the same platform also for automation across solutions. Finally, the highest level of transformation with AI, with full process redesign, will require an already good maturity of the enterprises, also in the areas of Data science, to develop a further level of value creation from the introduction of AI at a highly integrated process level. 
• As we raise the level of complexity of AI usage and tight integration with data, we see opportunities get stronger in terms of impact, but also reduce the number of enterprises capable of getting value from it and the effort to achieve, based on the actual business cases available on the market. One challenge I foresee, especially with lower maturity businesses, will be to invest a proportion of the IT budget in AI and its prerequisites, without eating part of the operational IT budget and granting those investments long enough to start to gain a consistent value creation increase. More mature enterprises are already structured to split those costs in the corresponding P&L at the right level of granularity to help to accelerate the most urgent priorities from business but the risk to have a benefit of AI in a specific department only or running from IT only without the proper business commitment, it’s a risk to consider because the most advanced digital transformations (including AI) are requiring long term focus and overall enterprise involvement.

Another separate topic is about what is also named AI Madness:

• One aspect of the limited or late regulations that we have run so far for the AI is that, for example, for some time, a big part of the training of AI has been done using content from the Internet not always freely accessible. Many content creators started to make noise about how AI could not just be trained with intellectual proprietary content. As a consequence, many contents have been built automatically with AI, especially in the area of translation. As a side effect, the need for a huge data source to train AI, tech AI companies went to use the same data that was already AI-generated as a source, causing a typical case of Model Autophagy disorder or MAD. In other words, systems risk getting less and less in the quality of answers because they train themselves using content from the Internet that was AI-generated as well. A quest for highly valuable data sources and intellectually protected content will definitely ramp up to not risk to collapse the whole engine.

There are many other interesting insights into the European future competitiveness, but one element I found relevant as a risk for Europe is the lack of quantum computing competencies. Out of the 10 top tech in quantum computing, 5 are in the US, 4 in China, and none in Europe. This is an area that I believe will ramp up strategically in the upcoming 5-10 years, especially linked to AI due to the capacity to use quantum computing capabilities to approach, for example, deep learning at next-level scale.

So correlating aspects, we see that some areas of the world have spent the last 20 years transforming more into digital businesses. There are no European businesses that grow from 0 to at least 100B$ in the last 50 years, but in the US, 6 enterprises grow in the same time from 0 to 1T$. All these US businesses are digital.

Those businesses built the infrastructure to scale AI, while other areas of the world remained on the more historical manufacturing model business, but started to spend a lot to regulate those new digital business models ramping up (for example, with the EU AI Act).

In the Social, we saw the following in September:

• It has also been the month of the sanction to Apple from the European Community for 13B€ due to an EU tax case from the past. 
• The remote work strategies made even less transparent the overall workforce allocation, as this accelerated dramatically during COVID. 
• Governments are adjusting rules, some enterprises are changing the initial plan partially, some, like Amazon, for example, announced they will be all back in the office 5 days a week for everyone from the 2nd of January 2025 (at least in the US, while in Europe some different rules will apply).
• These changes of work conditions and regulations evolving also on the right to disconnect for remote workers in EU, bring, from my perspective, a clear evolution on the attention for the wellbeing and employees relevant personal data to care that will drive the ramp-up of the evolution of human digital twin and the related HRIS and fitness capabilities, still keeping a clear consideration of the aspects exposed by the AI Act regulations (here I summarized key points) in term of privacy and emotions interpretation limitations applying on AI engines.

In the area of Environment, we saw:

• It’s interesting to see on one side the need for more power, more AI chipsets, computation, and data storage capabilities; on the other side, we know this is going to raise the level of energy needs. 
• It’s interesting that just in August I was referring in a post to a few risks I would see about discrepancies in consuming power needs for cloud datacenters and regulations restrictions in producing some type of energies in some countries. Europe is again a case where there is no common view about nuclear, for example.
• My reflection is questioning what will happen with the cost of cloud services in those countries that will require having data hosted locally and will have restrictions around the type of energy sources to operate? In my post, I was referring, for example, to Germany, which historically had the need to have within its borders cloud datacenters to guarantee strict data compliance, but at the same time decided to go against nuclear and has set a high bar for emissions.
• We just saw in September that Microsoft is restarting an old reactor in the US in Pennsylvania due to its AI power needs. Just for info, this is the plant that was the location of the most serious nuclear meltdown and radiation leak in US history.

On the chipset market, there are many transitions ongoing. Qualcomm and ARM still tried to scale up Intel to acquire their business, TSMC continues to grow its US production, and at the same time, some restructuring is happening, for example, in Intel and STMicro. The 2025 sees a big growth in the personal computers driven by the AI-powered new chipset, also listening to the opinion of big experts like Michael Dell. Definitely, an increase in the replacement of computers with AI-powered ones would bring more revenues for those vendors linked to that type of portfolio (Intel, AMD, Qualcomm), but they will have to effectively prove the value of AI-powered computers to push such an accelerated loop of renewal. Difficult to speculate, and is not my focus, on the market consolidation for chipset but definitely something could happen in the upcoming months.

One aspect where I see interesting possible future positioning for Europe, considering the high regulations around privacy, is about building proper services around highly privacy contents, for example, models for human digital twin with AI, and maybe use the HPC power from Europe to guarantee those data elaboration would remain within a certain highly regulated boundary. In this area where many psychological and body-sensory related information will be progressively stored and augmented to build a virtual view of people, there will come the need to have proper control over this information and the opportunity to interface with healthcare, insurance, pharma, and food enterprises, at least, to introduce capabilities bringing real, unique value. Here we speak about crossing data of health, food, body trends, wellbeing, remote and onsite work stress, and productivity together with gamification, prediction, and reduction of illness, general life quality improvement, and will bring improvement for all the parts involved. It’s definitely an opportunity that has to be explored from my point of view, on which EU already worked in definition and that has a big level of attention ramping up and I see opportunities to be integrated in the middle future with the full employees integrated wellbeing and HRIS strategies on one hand and the personal modern digital fitness augmented capabilities, health monitoring and avatar on the other.