Artificial intelligence has disrupted the 40-year cycle in the memory market: Micron's gross margin has reached 85%
The semiconductor memory market is undergoing a tectonic shift. The long-standing cycle that dictated the rules of the game in this segment for about four decades has been broken. The reason is the explosive demand from artificial intelligence. This is most clearly demonstrated by the earnings report of Micron Technology, which showed results unthinkable for this sector just a year ago.
The key indicator of the transformation is the company's gross margin. It soared to 85% compared to 39% a year earlier. For comparison: until recently, memory was considered one of the worst businesses in the technology sector, a cyclical commodity whose profitability fell to nearly zero every few years. Now, the most standardized product in the industry generates margins comparable to software companies, while remaining a physical device.
Micron's revenue quadrupled over the year, reaching $41 billion. Net profit jumped from $1.9 billion to $28 billion. But the main thing is the sustainability of the new position. The company's management stated that the market shortage will last at least until 2028. Half of future revenue is already secured in long-term contracts. A commodity does not receive such contracts — they are only given to a narrow link in the supply chain that is of critical importance.
The reason for this fundamental shift is not a change in the product itself, but a change in its value. Computing power for AI cannot function without memory, and there is a catastrophic shortage of it for everyone. It is the shortage, not the evolution of technology, that has turned the situation around. The cycle did not reverse — it broke.
Cryptalist Analytics: This event is not just a corporate success for Micron. It is a marker of how AI is reshaping fundamental economic models in the "hardware" sector. Investors should reconsider the valuation of companies producing "boring" components if they are on the critical path for AI infrastructure. Traditional cyclical metrics no longer work in this new context.