MiniMax president Yeyi Yun is trying to drag the AGI debate out of the seminar room and into the accounts. If AI can autonomously create about 1% of world GDP, she says, you can stop arguing about whether it counts.
Crypto Briefing reported that Yun, MiniMax's co-founder and president, gave that dollar-denominated test at the Goldman Sachs Asia Leadership Conference in Hong Kong on September 1. The number is not small. World Bank data published through the St. Louis Fed puts 2025 world GDP at $118.35 trillion, so 1% would be about $1.18 trillion. Use 2024 output and you're still talking about roughly $1.12 trillion.
That's a useful line. Most of the industry talks about artificial general intelligence in ways that are hard to check: OpenAI's charter refers to highly autonomous systems that outperform humans at most economically valuable work, the Turing test asks whether a machine can pass as human in conversation, and ARC-AGI measures abstract reasoning tasks. Yun went straight to the ledger. If the machines are not doing enough real work to show up in global output, don't bother calling it AGI.
She pointed to three early signals, according to the same report: autonomous planning, autonomous execution, and the ability of AI systems to check their own work without a human stepping in. None of that gets you to a trillion dollars by itself. It does tell you what MiniMax wants investors to watch. Not chatty demos. Not a benchmark screenshot. Actual work moving through actual systems.
The revenue behind the claim #
Yun's timing was not accidental. MiniMax had just reported first-half revenue of $116.6 million, up 283.1% from $30.4 million a year earlier, according to the company's August 26 interim results. Reuters also reported the same figure, noting that revenue from MiniMax's open platform and enterprise AI services rose 703.1% to $73.9 million. That business made up 63.4% of revenue in the first half, compared with 30.3% a year earlier.
MiniMax is bringing its AI listing story back to Shanghai MiniMax has started early preparations for a possible STAR Market listing in Shanghai after a sharp rally in its Hong Kong shares. The move would give the Chinese AI company deeper access to mainland capital while raising new questions about its global ambitions and compute strategy. - Chinese AI company MiniMax plans mainland listing strategy - how to do A-plus-H dual listing in China
The company is still spending heavily. MiniMax reported an adjusted net loss of $293 million for the six months ended June 30, versus $138.7 million a year earlier, while its cash balance stood at $1.32 billion. Growth is real here. So are the losses.
MiniMax has another number it wants you to notice: annualized recurring revenue passed $800 million in August, up from $150 million in February, according to Dealroom and the company's own results commentary. The company is targeting $1 billion in ARR by the end of 2026. That is the nearer test. A trillion-dollar AGI threshold makes for a sharp conference line, but $1 billion in run-rate revenue comes due much sooner.
The stock market has already shown how quickly it can reward that story. MiniMax listed in Hong Kong on January 9, and the South China Morning Post reported that its shares closed 109.1% above the IPO price, valuing the company at HK$106.7 billion, or about $13.7 billion. Forbes reported the same day that chairman and CEO Yan Junjie became a billionaire after the listing, with an estimated net worth of $3.2 billion. That was not a quiet debut.
The market is grading the AI labs now #
Z.AI shows the other side of the trade. The Hong Kong-listed company, known domestically as Zhipu AI, reported first-half revenue of 953.89 million yuan, about $142 million, on August 31. Bloomberg, carried by The Edge Singapore, reported that analysts had expected 1.35 billion yuan. Its loss narrowed to 2.07 billion yuan, but the revenue miss still exposed the problem sitting under China's AI boom: these companies can grow fast and still disappoint investors.
The comparison is useful because both companies are now public. You don't have to rely only on private-market mythology about model quality, user counts or who is closer to OpenAI. MiniMax and Z.AI are filing numbers every six months. Revenue mix, losses, cash, API growth, all of it is there for the market to punish or reward.
Frankly, that makes Yun's AGI definition more interesting than another philosophical line about machine intelligence. MiniMax is arguing that real AI progress should eventually be measurable in economic output, while its own business is being measured in much smaller but more immediate figures: $116.6 million in half-year revenue, $800 million in ARR, $293 million in adjusted net losses.
Is $1 billion in ARR by December realistic? MiniMax's run from $150 million to $800 million in six months says it might be. Its losses say the climb is not free. The bigger AGI argument can wait. For now, the company has four months to show whether the curve still holds.
Z.AI Missed Revenue Estimates by 29% Even as Its Valuation Hit $71 Billion Z.AI, the Hong Kong-listed maker of the GLM model family, reported first-half revenue of 954 million yuan ($142 million), missing Bloomberg's analyst estimate of 1.35 billion yuan by about 29%. The miss came even as revenue grew roughly 400% year over year, exposing how China's AI price war is punishing margins even at fast-growing labs like Z.AI... - how to beat China's AI price war competition - why AI startups miss revenue estimates despite growth
Also read: BLS Widens Its AI Job Risk List to Include PR, Web Design and Hotel Work • Baidu Says Its AI Business Is About to Earn as Much as Search Does • Wall Street Banks Are Pressing Law Firms to Cut Fees Because of AI