The Impact of Rising AI Computing Power Costs on the Industrial Chain

Upstream Hardware: A Super-Cycle for Chip Manufacturers

The price surge is driving industry prosperity across the entire chain. In upstream hardware, prices for server CPUs, AI chips, and memory devices are all rising, presenting historic opportunities for chip manufacturers.

NVIDIA is undoubtedly the biggest beneficiary of the current price surge. As the dominant player in global AI training chips, NVIDIA achieved significant revenue and profit growth in 2025. Demand for its latest Blackwell architecture chips far exceeds supply, with orders already scheduled into 2027.

Memory chip manufacturers are also reaping substantial benefits. HBM products from SK Hynix, Samsung Electronics, and Micron Technology remain in short supply, significantly boosting overall memory chip business profit margins. SK Hynix's 2025 financial report indicated that its HBM business revenue grew over 300% year-over-year, becoming the company's primary profit source.

Challengers like AMD and Intel are also actively positioning themselves in the AI chip market, seeking to capture a share of the growth. AMD's MI series chips have gained some market share in inference scenarios, while Intel is seeking breakthroughs in specific niches with its Gaudi product line.

Midstream Services: Strengthened Advantages for Leading Players
In the computing power service sector, the advantages of full-stack deployment for leading players continue to solidify. The three giants—Amazon AWS, Microsoft Azure, and Google Cloud—have further consolidated their market positions amidst the price surge, leveraging their scale advantages, technological expertise, and financial resources.

Following AWS's price increase for its Machine Learning Capacity Blocks, customer stickiness remains high. For enterprises that have already built AI applications on AWS, the migration cost far outweighs accepting the price increase. Similarly, Microsoft Azure has established a unique competitive advantage in AI services through its deep partnership with OpenAI.

In the 2025 global cloud infrastructure services market, AWS, Azure, and Google Cloud collectively held over 65% market share, a proportion that continues to grow slowly. Smaller cloud service providers face a more challenging situation amidst the price surge; they lack the bargaining power with upstream hardware manufacturers and struggle to leverage economies of scale to absorb costs.

End Applications: Cost Transmission and Innovation Opportunities
Rising computing power costs are being transmitted downstream. Price increases for API calls from large language model (LLM) vendors directly affect hundreds of thousands of AI application developers. Some small AI startups are facing cost pressures, forcing them to adjust their business models or seek funding.

However, the price surge is also catalyzing new business models. Some LLM vendors have introduced subscription-based packages, allowing enterprises to flexibly subscribe based on actual needs, ensuring token availability with predictable costs.

Concurrently, open-source models and smaller vendors are encountering opportunities for differentiated competition. As API prices from leading LLM providers rise, some price-sensitive users are turning to open-source models or services from smaller vendors, leading to more diversified choices in the market.