Over the past month, the market capitalization of leading chip manufacturers has lost more than $1.3 trillion. However, contrary to panic-driven sentiments, the root of the problem lies not in declining demand for semiconductors or technological lag. As an experienced analyst, I see that the market is misinterpreting the signals.
A telling example is Taiwan Semiconductor Manufacturing Co Ltd (TSMC). The company reported a record quarter: revenue reached $40.2 billion, and the growth forecast was raised above 40%. Profit surged by 77%, hitting an all-time high, and margins surpassed all previous peaks. TSMC also announced an additional $100 billion investment in its Arizona plants. These figures unequivocally indicate that demand for chips is not only not falling but is accelerating.
So what is the true reason for the sell-off? The problem has shifted from the silicon production sphere to the infrastructure sphere. Chips are no longer a scarce resource. Now, the development of the artificial intelligence (AI) industry is hitting a wall of steel, copper, and, critically, time.
Bottlenecks: From Packaging to Transformers
The first bottleneck is chip packaging. The CoWoS technology, which connects the processor to memory, is operating at its limit. Nvidia buys up about 60% of CoWoS capacity, and demand for this service has tripled in two years. TSMC is nearly doubling its production in this area, but the lines are still running at 100% capacity.
The second, and far more serious, constraint is energy. A finished chip is useless without electricity, and the industry is acutely short of it. Statistics on data centers in the US are alarming: companies have announced 16 gigawatts of capacity for 2026, but are only building 5 of them. The remaining projects are frozen, and 25% do not even have an energy supply plan.
The key element that the market has almost forgotten is the high-voltage transformer. It converts grid energy into power for the data center. Currently, its manufacturing takes an average of 48–60 months, whereas before 2020, 12 months sufficed. The shortage will persist for a long time: Hitachi, Siemens Energy, GE Vernova, and ABB — the four manufacturers of such units — have accepted orders years in advance. Siemens Energy alone has a portfolio volume reaching nearly 136 billion euros.
The problem is compounded at the raw material level. Special electrical steel for transformers is produced by only five companies in the world, and they cannot quickly ramp up production. Bypassing the shortage is also not feasible: some data center operators decided to supply their own energy and ordered gas turbines, but GE Vernova, Siemens, and Mitsubishi have already sold their capacity for about five years ahead.
Analyst's Conclusion: Recalculating Substations
The situation fundamentally changes the view on the sell-offs. Traders were dumping stocks due to fear of falling demand, but the data points to a supply shortage — not in chips, but in infrastructure. Beneficiaries could include transformer and turbine manufacturers, grid companies, and owners of sites with ready energy connections.
My professional opinion: The market recalculated chips, when it should have recalculated substations. The AI story is now not about intelligence, but about electricity. The world's most advanced companies are forced to build their own power plants to run neural networks. Investors should shift their focus from semiconductor giants to "concrete-and-steel" energy infrastructure — that is where a new cycle of super-profits is currently forming.