Artificial general intelligence (AGI) refers to a theoretical area of AI research aimed at creating software that emulates human-like intelligence and can learn autonomously. The goal is for this software to execute tasks it is not specifically trained or designed to perform, according to AWS.
He mentioned that the firm intends to keep its most sophisticated AI models open-source, contending that open-source development and commercial profit generation can coexist.
DeepSeek attracted worldwide attention early last year when its affordable V3 and R1 models challenged perceptions of China’s competitiveness against US AI powerhouses, despite Washington’s restrictions on advanced chip exports.
The company has recently concluded its first fundraising round with a valuation of approximately $52 billion, signaling a departure from its previous approach of rejecting outside investment. This change comes amid intensifying competition from local challengers like ByteDance, Moonshot, and Z.ai.
Liang indicated that the primary challenge for the company is access to computational power, emphasizing that the major gap with the US lies more in resource availability than in technical expertise.
He added that DeepSeek plans to construct its own large AI computing clusters in the future, though a decision on developing its own AI chips has yet to be made. “I hope to be able to purchase chips at reasonable prices so that I don’t have to manufacture them myself,” Liang stated.
However, the Reuters report highlighted that while the US has imposed restrictions on exporting advanced AI chips to China, it has not added DeepSeek, the Chinese memory chipmaker CXMT, and over 100 other firms with national security concerns to a trade blacklist, aiming to prevent escalating tensions with Beijing.
Also Read: China’s DeepSeek developing its own AI chip, sources say
