%20transparent.png){kind=small}
T&V Letter | Q2 2026
- Thalmann & Verling Trust reg.
- 4 days ago
- 5 min read
Insight
In Japan, according to market leader Unicharm, more adult diapers have been sold than baby diapers since as early as 2011. What sounds like a quirky footnote is in truth an early warning sign. Japan is anticipating demographic developments that will reach us too. Already today, more than one in ten people there is over 80 years old, nearly 30 percent are over 65, and entire regions are depopulating because so few children are being born.
Switzerland has reached similar turning points as well. According to the Federal Statistical Office, in 2025 the number of people aged 65 and over exceeded that of people under 20 for the first time. Since 2019, more people have also been retiring each year than young workers entering the workforce. As a result, around 30,000 people a year are missing who would otherwise be paying into old-age and pension funds or paying taxes. Another figure shows just how much the country is changing: around 1,500 people over the age of 100 currently live in Switzerland, and by 2050 that number is expected to reach roughly 15,000.
The trigger for this development is the birth rate. The most recent available figures show a value of 1.29 children per woman for 2024, the lowest level since records began. A stable population without immigration would require 2.1 children per woman. Switzerland is not an isolated case. A study published in the scientific journal The Lancet estimates that by the year 2100, only six of roughly 200 countries will still have a birth rate above the replacement level.
For the economy, this is of major significance. An aging society changes everything, from social insurance systems to the labor market, property prices, and the question of which industries will grow and which will shrink. Precisely for this reason, it is worth keeping an eye on this shift. Because it unfolds gradually, it is still barely noticed by many.
Dear reader
Stock markets have gained significantly since the start of the year. The leading US index, the S&P 500, repeatedly reached new all-time highs, driven above all by major technology companies linked to artificial intelligence.
Anyone wanting to understand this development should not be guided by headlines, but should look at the facts. Contrary to what is often assumed, this development is not carried by euphoria alone, but by strongly rising earnings. In the first quarter, earnings of S&P 500 companies rose by around 21 percent. A large share of this growth comes from companies linked to artificial intelligence. Because earnings grew faster than prices, US stocks are actually somewhat more attractively valued than before, not more expensive as many assume. From an earnings standpoint, markets are therefore not overvalued, though it remains to be seen whether this earnings growth can be sustained at these levels.
It is also striking how calmly markets are currently handling a challenging macroeconomic environment: inflation, budget deficits, tariffs, and higher interest rates are weighing on them only slightly. On one of these fronts, there was finally some relief recently. In mid-June, President Trump announced a preliminary agreement with Iran on reopening the Strait of Hormuz, through which around a fifth of the world's oil flows. The previously sharply risen oil price then fell significantly.
But while many are watching oil, a more fundamental development is taking shape in another commodity. One that powers our modern world: electricity.
Electricity Becomes the New Oil
For decades, electricity in industrialized countries was a topic for engineers, not investors. More efficient devices offset rising demand, and in Europe and the US demand stagnated for over twenty years. That era is over, and the world's largest technology companies are responsible for it. For artificial intelligence, Microsoft, Amazon, Alphabet, Meta, and Oracle, known in the trade as “hyperscalers”, are building enormous data centers and expanding their capacity at a breathtaking pace.
These capacities are needed above all for training new AI models, a process that consumes enormous amounts of energy. AI thought leader Leopold Aschenbrenner describes, in his widely noted essay written back in 2024, how these companies have entered a race for ever larger compute clusters, whose power requirements exceed every previous order of magnitude. The cluster on which OpenAI trained the GPT-4 model in 2022 required around 10 MW, roughly as much as ten thousand households, and cost close to 500 million dollars. Aschenbrenner wrote two years ago that the power and cost of these facilities roughly increase tenfold every two years. For 2026, this already implied a cluster of around 1 GW, with costs in the tens of billions of dollars.
How close this forecast is to reality is already becoming clear today. In June 2026, Microsoft and Chevron announced a data center in Pecos, Texas, whose power demand at full build-out is expected to reach around 2.67 GW. Power will be supplied by a dedicated gas-fired power plant with seven large gas turbines, entirely independent of the public grid. For reference: one gigawatt is roughly equivalent to the output of a nuclear power plant. A large gas turbine delivers around 400 MW, a modern wind turbine only about 6 MW. This data center therefore consumes almost as much electricity as three nuclear power plants, and does so around the clock.
And according to Aschenbrenner, this development is only just beginning. For 2028, he expects clusters of 10 GW and costs in the hundreds of billions of dollars, equivalent in energy terms to the consumption of a smaller US state. By 2030, he considers a single cluster of 100 GW possible, with costs exceeding one trillion dollars, which alone would claim more than twenty percent of the entire current US electricity production.
The Real Bottleneck
Demand for electricity is growing far faster than supply. Since 1985, US electricity production has increased by only around 60 percent. This sluggish supply side is now meeting demand that, driven by artificial intelligence, is reaching entirely new orders of magnitude. According to Aschenbrenner, investment in AI infrastructure is set to multiply from around 150 billion dollars in 2024 to roughly 8’000 billion dollars by 2030, an increase of more than fiftyfold. For 2026, he estimated around 500 billion dollars. In fact, the announced investment total from hyperscalers for 2026 now stands at around 725 billion dollars, already 45 percent above his estimate. These enormous sums translate into revenue for suppliers. Their sharply rising profits are one of the drivers behind the index earnings growth mentioned at the outset.
Where the energy is supposed to come from is therefore not an easy question. Solar and wind are inexpensive but do not supply power around the clock. In the short term, natural gas above all is stepping in, and nuclear power is also making a comeback. Operators are also increasingly turning to fuel cells that can be installed directly on site. All these solutions share one goal: avoiding the years-long wait for a grid connection by having power generation arise directly next to the data center, as in the Texas project mentioned above.
Our Conclusion
Electricity is becoming a strategic resource, much like oil in the 20th century. Already today, Gulf states are luring hyperscalers with cheap, near-unlimited energy. At the same time, voices like Aschenbrenner's warn against leaving this key infrastructure to authoritarian regimes. Because whoever controls the facilities on which the most advanced AI is built holds a strategic lever of power of the first order. What matters is not only who builds the best models, but who owns the energy, the data centers, and the control.
As always, we thank you for the trust you have placed in us!
*This communication is for information purposes only and constitutes neither a personal recommendation nor an independent financial analysis.
Comments