In 2025, major technology companies, including Microsoft, Amazon, and Meta, are expected to invest 371 billion dollars in AI Datacenters and computing resources, a 44% increase from the previous year. By 2032, this figure will rise to 525 billion dollars, with inference-driven investments potentially accounting for nearly half of all AI spending that year.
The rise of the Chinese company DeepSeek is reshaping the Global AI investment landscape, forcing USA technology companies to reassess their AI strategy. Driven by DeepSeek and OpenAI's new reasoning models, tech giants are expected to increase annual AI investments to 525 billion USD by 2032, with a shift in focus from AI model training to reasoning capabilities.
On March 17, Bloomberg reported that Global technology giants are preparing to significantly increase AI investments, by 2025, including $Microsoft (MSFT.US)$ 、 $Amazon (AMZN.US)$ and $Meta Platforms (META.US)$ Tech giants, including others, are expected to invest $371 billion in AI Datacenters and computing resources, a 44% increase from the previous year. By 2032, this figure is projected to rise to $525 billion, a growth rate far exceeding Bloomberg Intelligence's expectations before DeepSeek gained popularity.
The rise of the Chinese AI startup DeepSeek has had a strong impact on the USA Technology Industry, as the company claims to have developed competitive models at costs far lower than those of leading competitors in the USA, raising doubts within the USA Technology Industry about its heavy investments in AI development strategy; some leading AI companies are now beginning to embrace efficient AI systems that can run on fewer chips.
Bloomberg Analyst Mandeep Singh noted in the report:
"Capital expenditures on AI training may grow far less than we previously expected. The significant attention garnered by DeepSeek may drive Technology companies to increase investments in reasoning capabilities, making it the fastest-growing area of the generative AI market."
The shift in investment focus is particularly evident. While spending related to training is expected to account for over 40% of large companies' AI budgets this year, this proportion is expected to drop to only 14% by 2032. In contrast, reasoning-driven investments could account for nearly half of all AI spending that year.
This strategic shift will reshape the industry landscape. Some analysts point out that Google, with its self-developed chips, seems most capable of quickly completing this transition, as its internal chips can handle both training and reasoning tasks simultaneously. In contrast, companies such as Microsoft and Meta Platforms, which rely heavily on $NVIDIA (NVDA.US)$ chips may lack flexibility and face greater challenges in transformation.
Currently, technology giants are accelerating the shift of investment focus from AI model training to inference — the process of running these systems once model training is complete. These systems require more time to compute user query responses, simulating the human thought process, but they also provide new opportunities for software revenue and may shift some costs from the development phase to after the model launch.
Editor/rice
Comment(3)
Reason For Report