The World Artificial Intelligence Conference (WAIC) opened in Shanghai on Friday. The SCMP has covered WAIC since its inception in 2018, when our story headline was: “China wants to be the world leader in AI. Here’s how they plan to pull it off”.

What was considered state of the art in AI nearly a decade ago? 

Our 2018 report highlighted medical image analysis, voice assistants and smart speakers from Tencent, Xiaomi and Baidu, respectively.

Pretty primitive by today’s AI standards. This year's event features a host of AI agents and a thought-controlled robot.

The “World” part of WAIC is now somewhat of a misnomer, as post-Covid it has largely been a domestic affair without headlining foreign company leaders like Elon Musk, who debated Alibaba founder Jack Ma on the future of AI and humanity on stage in 2019. (Disclaimer: Alibaba owns SCMP).

WAIC provides Beijing an opportunity to press its claim to a larger role in shaping global AI governance. A keynote speech by Chinese President Xi Jinping at the opening ceremony on Friday elevated the conference from an industry showcase to a platform for national technology strategy and diplomacy. 

Stay tuned over the weekend for more SCMP reports from WAIC.

AGI or bust

AGI or bust

Clearly, the stakes in the US-China AI race have escalated dramatically since 2018. “Winning” today is usually interpreted as the first to develop artificial general intelligence (AGI). 

The CEOs of two major Chinese AI model developers recently declared their intention to focus on pursuing AGI over short-term profit gains.

Zhipu AI (which trades as Knowledge Atlas Technology in Hong Kong), said it would spend the next two years investing in frontier AI research, according to an internal letter issued by co-founder Tang Jie. 

The company “would not pursue short-term monetization from AI applications”, but instead set its sights on the next frontier of AGI, when AI captures “the aggregation of all human intelligence”, Tang said in the letter.

Earlier, the founder and CEO of rival MiniMax, Yan Junjie, expressed his determination to achieve AGI and pledged to forgo his salary until that level of intelligence was realized. 

In contrast to those pursuits, two former Chinese AI lab leaders said they aimed to focus on building industry-specific AI solutions and models following their exit from the race for AGI.

“All models are relentlessly chasing the ceiling of general intelligence … but in many real-world scenarios, we don’t necessarily need an absolute all-rounder,” said Zhang Fan, founder and CEO of Yoolee AI and previously chief operating officer at Zhipu AI, in an interview with the SCMP.

Zhang said that while current AI models were “extremely powerful”, few major enterprises had seen AI generate tangible business value.

Another interesting trend in AI is the emergence of the “token economy”. As tech giants rush to encourage staff to integrate AI into daily workflows, tokens – the basic unit through which AI services are measured and billed – have become a new kind of corporate allowance. 

“The [Chinese] digital economy has gone through the stages of the data economy and the computing economy. Today, the token economy is emerging,” said Yin Hao, an academic at the Chinese Academy of Sciences.

Who knows, maybe one day our salaries will be in AI tokens? 

Priceless memories

Priceless memories

AI tokens aren’t just changing how corporations do business. The massive explosion in token usage has fundamentally changed the dynamics of the memory chip industry. 

For many decades, the DRAM industry suffered cyclical downturns every two to four years when market demand for computers would stall. Chip equipment makers, in particular, were hit hard. 

Teradyne co-founder Alex D’Arbelloff once told me: “We had been growing very, very rapidly through the '60s, almost doubling every year, and then came the recession of 1970. For three months starting in mid ‘70 we didn't get a single order. It was really a shock, the first of the cyclical downturns.”

Today, nobody is forecasting a cyclical downturn in memory anytime soon. It’s a real eye-opener to see the share price history of Micron Technology (DRAMs) and Teradyne (test gear) over the past five years. Both languished until a year ago, and have since skyrocketed.

Chinese memory makers are also taking advantage of the newfound investor fervor for semiconductor memory. 

More than 9.4 million investor accounts applied for shares in the blockbuster initial public offering (IPO) of ChangXin Memory Technologies (CXMT), with retail demand reaching about 212 times the shares initially available online.

“For A-share IPOs in China, winning an allocation [usually] means making money,” said Kevin Chen, a 35-year-old employee at a state-owned enterprise in Shanghai, who also subscribed for shares.

Professional investors also showed strong interest. Among them,153 private fund products managed by High-Flyer Quant, the hedge fund co-founded by DeepSeek founder Liang Wenfeng, submitted bids for a combined 12.55 billion shares.

The Hefei-based CXMT is seeking to challenge Samsung Electronics, SK Hynix and Micron. The Chinese company currently holds a 9 per cent global share of DRAMs measured by bits, according to Counterpoint Research. 

That’s all for this edition of China Future Tech. Stay tuned for more AI and chip news from China.

The best of SCMP Tech