Chinese AI rival spending 2% of ChatGPT’s budget – industry pioneer

Must read

REVIEW: Impact of Evidence by Carol Carnac

JayneBook Reviews / C- Reviews1950s / Classic Murder Mystery / England / English countryside / murder mystery / Post WWIINo CommentsNear St. Brynneys in...

'Rich Dad, Poor Dad' author calls for $1 million BTC by 2035

Financial educator, author of Rich Dad, Poor Dad, and investor Robert Kiyosaki recently forecasted a $1 million Bitcoin (BTC) price by 2035 as the...

Charles Schwab CEO eyes spot Bitcoin trading by April 2026

Schwab remains cautious yet optimistic about the digital asset sector and is looking to add spot BTC trading services for its clients. ...

Crypto industry is not experiencing regulatory capture — Attorney

Transparency, panel hearings, and public input are critical for healthy crypto regulations, attorney Brandon Ferrick told Cointelegraph. 2239 Total views ...

Kai-Fu Lee has cast doubt on the sustainability of OpenAI’s model compared to open-source DeepSeek

An artificial intelligence pioneer has questioned the sustainability of ChatGPT creator OpenAI, noting that Chinese rival DeepSeek has been trained at a fraction of the cost.

DeepSeek heated up an AI race between China and the US in January after launching an open-source version of its reasoning AI model, R1. Unlike its American rival ChatGPT, the Chinese bot is freely available without a subscription, which has helped it rapidly become the most downloaded app on Apple’s and Google’s stores.

“If you think about OpenAI’s $7 billion of operating costs in 2024, DeepSeek is probably operating with 2% of the operating expense,” Kai-Fu Lee, a writer, venture capitalist, and technology executive who has over 30 years’ experience in AI, said on Thursday in an interview with Bloomberg.

Lee is the founder of Sinovation Ventures, a venture capital firm focused on supporting Chinese startups in AI and technology, as well as establishing an AI institute which focuses on integrating new technology into traditional industries and developing AI-driven startups. He previously served as founding director of Microsoft Research Asia before working at Google and Apple.

“Is OpenAI’s model even sustainable?” the expert asked.

Lee specified that OpenAI was spending $87 billion a year “making a massive loss.” He compared this to DeepSeek, which is offered to developers for free or at a fraction of the cost.

“So with that kind of formidable competitor, I think Sam Altman is probably not sleeping well,” he said, referring to the OpenAI CEO. 

The venture investor and founder of 01.AI projected that only three AI models – DeepSeek, Alibaba Group Holding, and ByteDance – would survive China’s domestic shakeup triggered by the onset of DeepSeek. Lee added that xAI, OpenAI, Google, and Anthropic could lead the sector in the US.

He emphasized that pioneer models such as ChatGPT have been commoditized, making the business model of the US developer vulnerable to cheaper economics presented by open-source AI such as DeepSeek.

More articles

Latest article

REVIEW: Impact of Evidence by Carol Carnac

JayneBook Reviews / C- Reviews1950s / Classic Murder Mystery / England / English countryside / murder mystery / Post WWIINo CommentsNear St. Brynneys in...

'Rich Dad, Poor Dad' author calls for $1 million BTC by 2035

Financial educator, author of Rich Dad, Poor Dad, and investor Robert Kiyosaki recently forecasted a $1 million Bitcoin (BTC) price by 2035 as the...

Charles Schwab CEO eyes spot Bitcoin trading by April 2026

Schwab remains cautious yet optimistic about the digital asset sector and is looking to add spot BTC trading services for its clients. ...

Crypto industry is not experiencing regulatory capture — Attorney

Transparency, panel hearings, and public input are critical for healthy crypto regulations, attorney Brandon Ferrick told Cointelegraph. 2239 Total views ...

Every chain is an island: crypto’s liquidity crisis

Opinion by: Jin Kwon, co-founder and chief strategy officer at SagaCrypto has come a long way in boosting transaction throughput. New layer 1s (L1s)...