The explosive growth of artificial intelligence is generating enormous profits for memory chip manufacturers, but those gains are coming at the expense of AI developers, cloud providers, and potentially consumers. Companies such as Micron Technology, Samsung Electronics, and SK Hynix have benefited from surging demand for high-bandwidth memory (HBM), an essential component for training and running advanced AI models. With manufacturing capacity constrained and new fabrication plants taking years to build, memory chip prices have risen sharply, shifting profits toward chip suppliers.
According to the article, prices for DRAM and NAND memory have increased dramatically in recent months, significantly raising infrastructure costs for AI companies. Since leading AI providers such as OpenAI and Anthropic continue to price their services to attract users rather than maximize profits, they are absorbing much of these higher costs instead of passing them on to customers. This has reduced margins for AI model developers and data center operators while boosting earnings for chip manufacturers.
The article also notes that enterprises are becoming more cautious about AI spending. As employee usage of AI tools grows rapidly, some organizations have begun rationing AI usage because increased token consumption has not always translated into proportional productivity gains. Raising AI service prices to offset higher chip costs could further slow enterprise adoption, making profitability an even greater challenge for AI providers.
The article concludes that the AI boom is causing a major redistribution of profits across the technology supply chain. While semiconductor companies currently capture a growing share of the economic value created by AI, model developers and cloud providers face mounting cost pressures. Until chip supply expands or AI services become consistently profitable, the financial benefits of the AI revolution are likely to remain concentrated among companies that manufacture the critical hardware powering it.
