Blackstone defends data centre investment vision amid DeepSeek AI disruption
Written by Nicole Cappella Thu 6 Feb 2025
Blackstone adopted a defensive posture during its most recent earnings call, responding to increasing worries about the future of large-scale data centre investments following DeepSeek’s recently revealed advancements in AI efficiency.
The Chinese company’s innovative approach to AI has led many to question the continued need for enormous and ever-expanding data centres, sparking volatility in tech stocks.
Senior research and development manager of data science at Black Duck, Dr Andrew Bolster, commented that the release of DeepSeek ‘undeniably showcases the immense potential of open-source AI’.
“Open-source AI, with its transparency and collective development, often outpaces closed-source alternatives in terms of adaptability and trust. As more organisations recognise these benefits, we could indeed see a significant shift towards open-source AI, driving a new era of technological advancement,” added Bolster.
Blackstone failed to address its AI data centre strategy at the start of the earnings call, but responded to a question from Morgan Stanley analysts, who pressed Blackstone on the implications of DeepSeek’s advancements.
COO of Blackstone, Jon Gray, acknowledged the potential of open-source AI to shift demand but emphasised the firm’s confidence in its data centre strategy.
Blackstone Bets Big on AI
This discussion followed Blackstone’s July 2024 earnings call, which was largely focused on the company’s AI-driven data centre strategy. At the time, CEO Stephen Schwarzman projected around £804 billion ($1 trillion) in capital expenditures for US data centre development over the next five years, with another £804 billion ($1 trillion) internationally. He also noted the demand for electricity in the US could rise by 40% due to the power requirements of AI-driven data centres.
Gray acknowledged the change that DeepSeek could press on the data centre industry but remained confident that other factors will still contribute to ongoing data centre demand.
“The cost of compute is coming down pretty dramatically … but at the same time, that is going to lead to more usage and more adoption,” said Gray. Increases in inference workloads, cloud services, and enterprise applications will offset a decrease in demand for training models and processes.
“We still think it is a very important segment, and there is s a way to run … But obviously, we are watching what is happening very closely,” added Gray.
Written by Nicole Cappella Thu 6 Feb 2025