AI Firm Plans Private Credit Fund for Servers With Nvidia Chips

June 3, 2024

28 1 minute read

Untitled design6 — eb358177f75cd234b13a804b5753912d.jpeg

(Bloomberg) — A British artificial intelligence startup plans to launch a private credit fund to buy servers outfitted with high-powered Nvidia Corp. chips to be used by companies in Asia.

Most Read from Bloomberg

In its first foray into private credit, Venture Hub, a UK-based AI company operated by ANC Research & Development Ltd., seeks to raise up to $50 million to buy servers using Nvidia H200 graphic processing units, director and founder Alex Cheung told Bloomberg News in an interview Monday.

The goal is to establish an initial cluster of at least 96 servers with H200 GPUs. The H200 is Nvidia’s latest GPU and has capabilities that allow it to better cope with large data sets needed for AI, the company has said.

The fund’s launch plan comes as Asia faces a shortage of computing power for high-end technology processes, while investment is flowing into for data centers as demand for AI capabilities grows. The world’s biggest companies are set to spend up to $60 billion over the next couple of years on data centers in Southeast Asia as the region embraces video streaming, online shopping and generative AI.

Clients would include companies based in the Asia-Pacific and as well as government or related departments in Taiwan, according to Cheung. Revenue would be generated by charging an hourly fee to access the servers with the chips, Cheung added. Venture Hub was founded in late 2022 and currently has six employees, including Cheung.

“The servers and the Nvidia chips depreciate with time,” said Cheung. “That’s why investors are keen to use private credit as an investment strategy: It’s the most practical way to enjoy the short-term returns while not having to buy the asset themselves,” he added.

Venture Hub has reached out to family offices in Hong Kong, Singapore and Taiwan with a number expressing interest, Cheung said.

H200 GPUs can help data centers accelerate processes including machine learning and high-performance computing, Nvidia has said — technology needed to handle reams of data faster.

The earliest expected launch date for the fund is end of August, Cheung said.