Interpretation of Bain's 2025 Technology Report: The demand for AI computing power is growing far beyond Moore's Law, and the transformation of enterprises' "core IT infrastructure" is imperative
The Bain report shows that global AI computing power demand is growing at a rate of 4.5 times per year, far exceeding the efficiency improvement rate of doubling every two years as per Moore's Law. To meet this demand, an additional 200 gigawatts of computing power capacity will be needed globally by 2030. If companies want to truly profit from advanced AI intelligences, they must modernize their outdated core IT architectures
Author: Long Yue
Source: Hard AI
A technology report from Bain for 2025 reveals a harsh and clear reality behind the AI revolution: the demand for computing power is growing out of control, while corporate infrastructure is severely lagging behind.
On October 20, Barclays analyst Raimo Lenschow analyzed this report, believing it reveals two irreversible trends: The growth rate of AI's demand for computing power has far exceeded Moore's Law, and if companies want to profit from it, they must modernize their outdated core IT architecture.
Computing Power Demand Far Exceeds Moore's Law
The most striking point of the report is that AI-driven demand for computing power is growing at a rate of 4.5 times per year, which is more than twice the speed of Moore's Law (which states that chip efficiency roughly doubles every two years).
This means that technological innovation can no longer keep pace with demand. This imbalance will trigger massive investments in data center infrastructure.
According to Bain's analysis, by 2030, global computing power demand could reach 200 gigawatts (GW), with the United States alone accounting for 100 GW. To build such capacity, approximately $500 billion in capital expenditure will be needed each year, corresponding to $2 trillion in annual revenue.
Barclays points out that although government subsidies are unlikely to fill such a huge funding gap, actions from the private sector are more aggressive. Moves like OpenAI signing trillion-dollar-level computing contracts indicate that future investment levels will only continue to rise.
This prediction is being confirmed by market actions. According to Reuters, an investment group supported by BlackRock and NVIDIA recently acquired data center operator Aligned Data Centers, which has nearly 80 sites, for $40 billion. Meanwhile, Morgan Stanley estimates that tech giants including Alphabet, Amazon, and Meta will spend $400 billion on AI infrastructure this year.
Corporate Modernization: The Only Path to Harness AI
Beyond the computing power race, another core aspect of the report is that corporate IT modernization is urgent.
As AI evolves from simple information retrieval (first-level agents) to executing complex workflows (second and third-level agents), companies find themselves unprepared. Corporate clients remain cautious about fully relying on autonomous AI agents without human oversight.
To truly leverage AI, especially more intelligent "Agentic AI," companies must undergo thorough IT modernization. Agentic AI can think, reason, and act based on user instructions, far exceeding simple information retrieval chatbots.
Bain believes the solution lies in changing the architecture around the agents. The report suggests, "Companies must modernize their core IT infrastructure and build in observability features to provide a suitable ecosystem for these agents."
For investors, this means that in the second half of the AI game, the battlefield will shift from purely computing hardware to software and service providers that can help companies complete this painful but necessary transformation. This is not just a technological upgrade, but a battle for survival This article is from the WeChat public account "Hard AI". For more cutting-edge AI news, please click here
