Market sentiment around artificial intelligence continues to swing between fear and euphoria on a weekly basis. Headlines shift from “bubble” to “breakthrough” within days of each other. We prefer to look past the noise and focus on the underlying facts, because the data tells a more interesting story than the mood swings do.
The shortage of memory chips is intensifying, and it is becoming the defining constraint of this AI cycle. Nvidia has now confirmed SK Hynix, Samsung, and Micron as the three suppliers for the next generation of HBM memory. Fujikura, the Japanese fiber-optic cable manufacturer, was forced to raise its profit forecast on the back of surging demand. Micron posted record results, with revenue growth of roughly 350%. HD Hyundai Electric, meanwhile, secured a major contract from a U.S. hyperscaler worth close to a quarter of its annual revenue, a sign of just how much capital is flowing into the physical infrastructure behind AI.
Where is this heading?
Demand for AI services has risen sharply and now stands at around USD 70 billion a year, served primarily by three leading providers: Anthropic, Gemini, and OpenAI. We don’t rule out this market expanding, within a few years, to represent several percentage points of global GDP — several trillion U.S. dollars annually. That would make it one of the fastest-scaling technology markets in history.
At the same time, the key bottleneck is becoming clear. More than half of the investment growth by hyperscalers in large-scale data centers is attributable to rising memory chip costs, meaning actual capacity expansion is progressing more slowly than headline capex figures suggest. In other words, the dollars going in are growing faster than the compute coming out. The growing complexity of AI models, combined with the rising share of inference relative to training, is sustaining the DRAM shortage, one we expect to persist for years rather than quarters.
Top contributors
Tesla’s news flow remains positive. An increasing number of countries are authorizing the use of its self-driving technology, a trend now gradually extending across Europe as well. We view this segment — not the car business itself — as the company’s most valuable asset. By contrast, we deliberately chose not to participate in the SpaceX IPO. In our view, the probability of success is too limited relative to the valuation, and we expect the company to require substantial amounts of additional capital in the coming years.
Frank Schwarz recently traveled to New York for a series of company meetings, including with Axon Enterprise, one of our most promising software holdings. Artificial intelligence is creating additional growth opportunities for the company: demand for Draft One, the AI agent used by law enforcement agencies to automatically generate incident reports, is expanding rapidly. Axon also stands out for its strong customer retention and revenue growth of more than 30%. The stock was among the largest positive contributors to fund performance this month, gaining 27%.
Detractors
Not every part of the portfolio pulled its weight this month. Gold, silver, and copper prices came under pressure, weighing on our Resources investment theme. Sentiment among American consumers also remains fragile: U.S. restaurant chains are feeling the impact of weakening purchasing power, and household spending continues to depend heavily on affordable energy prices.
The Chinese automotive market shows little sign of improvement for the time being. Internal combustion engine vehicles are currently being sold at average discounts of approximately 25%, a level of price competition that is squeezing margins across the board. This trend is reflected in BMW’s recent profit warning: the company now expects its 2026 operating margin to decline from 6% to 1%, with China cited as the primary driver behind the deterioration. Whether conditions will improve in the near term remains uncertain.
Portfolio changes
We initiated a new position in ASE Technology. The Taiwanese company benefits from the increasing complexity of packaging and testing requirements for modern semiconductor generations and stands out for its rapidly improving revenue and earnings expectations. It is a direct beneficiary, in our view, of the same memory-driven bottleneck reshaping the rest of the sector.
Within the consumer sector, we invested in Fast Retailing. Best known for its Uniqlo brand, the Japanese company continues to benefit from its international expansion, particularly in the United States and Europe, while delivering revenue growth of approximately 15%.
Riding the Tech Wave