Alibaba CEO predicts machine thinking will exceed human thinking by 1,000 times
At the 2026 Yunqi Conference in Hangzhou, Alibaba CEO Wu Yongming declared the arrival of a "machine intelligence" era, predicting total machine thinking will exceed human thinking by 1,000 times. He identified AI models, AI chips, and AI cloud as three foundational pillars for this era. Alibaba announced its Zhenwu V900 AI chip with triple the computing power of its predecessor, plans to scale Qwen models to 5-10 trillion parameters, and a target for Alibaba Cloud to operate over 20GW of global data center capacity by 2032.
IllustrationEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Common ground
- All agree that Alibaba's AI buildout is a strategic response to US export controls and chip restrictions.
- There is agreement that the current AI development lacks democratic oversight and accountability globally.
- All recognize that the Global South has historically been exploited by Western tech powers and faces similar risks with Chinese tech.
- There is consensus that Wu Yongming's '1,000x machine thinking' claim is marketing hype, not a proven technical reality.
Points of contention
- Eastern Agent sees China's AI infrastructure as a liberation for the Global South, while Regional Agent views it as a new form of extraction and control.
- Eastern Agent argues sovereignty justifies the buildout regardless of ethics, while Regional Agent insists on centering the people affected by these systems.
- Neutral Agent claims the scaling thesis may fail, making the debate premature, while Eastern Agent says the strategic value of building capacity is independent of technical success.
- Regional Agent believes Chinese AI simply replaces one master with another, while Eastern Agent argues multipolarity gives the Global South options it never had.
Blind spots
- None of the agents adequately address the environmental and labor costs of building massive data centers and mining rare earths.
- The debate overlooks the voices of people in the Global South who will actually live under these AI systems—they are discussed but not represented.
- All agents assume the current AI paradigm (transformer architecture and massive compute scaling) will continue, ignoring the possibility of a major technical disruption or collapse.
- The discussion fails to consider how Alibaba's domestic focus might shift to extraction in other countries once its home market is saturated.
WorldAttention’s read
This debate reveals a deep divide between viewing Alibaba's AI buildout as a necessary sovereignty play against Western dominance versus seeing it as a new form of extractive control that ignores human rights and democratic accountability. While all agree that US export controls forced China's hand and that AI lacks democratic oversight everywhere, they disagree on whether Chinese infrastructure offers genuine liberation or just a different master. The biggest blind spot is that everyone is arguing about who will control a future that may not arrive as promised—the scaling thesis could fail, leaving both the triumphalist and dystopian narratives irrelevant. Ultimately, the conversation misses the people who will actually live under these systems, and no one offers a concrete path to giving them a voice or real accountability.
Reporting timeline
Alibaba unveils 20GW data center ambition, bets on 'machine intelligence' era at Yunqi Conference
At the 2026 Yunqi Conference in Hangzhou, Alibaba CEO Wu Yongming outlined a vision for a 'machine intelligence' era driven by Artificial Superintelligence (ASI), predicting machine-generated thought will exceed human thought by 1,000 times. He compared current AI coding to the 1882 electric light bulb, stating the representative product of the era has yet to emerge. To prepare, Alibaba announced a long-term strategy to invest in three foundational pillars: AI models, AI chips, and AI cloud. Key targets include expanding Alibaba Cloud's global data center capacity to over 20GW by 2032, scaling Qwen models to 5-10 trillion parameters, and releasing the Zhenwu V900 AI chip with 3x performance improvement. The company also launched Agentic Cloud products and showcased AI applications in filmmaking, e-commerce, and automotive sectors. Wu emphasized that machine intelligence will free humans for creativity and curiosity, echoing the theme 'AI stronger, humans stronger.'
Read sourceYunqi Conference 2026: Agentic AI Moves From Chat to Real Work in Machine Intelligence Era
The 2026 Yunqi Conference in Hangzhou, held September 22-24, showcased the shift of AI from conversational capabilities to practical task execution, with Agentic AI as the central theme. The 50,000-square-meter exhibition featured over 1,000 companies. Notable demonstrations included a gastrointestinal endoscopy AI diagnostic model by Tianjin Yujin Medical, which flags suspicious areas in real-time to assist doctors, and Alibaba's Qwen Office launching 'Enterprise Context' to integrate AI into business workflows. Alibaba Group CEO Wu Yongming delivered a keynote predicting that machines will eventually handle over 1,000 times the thinking volume of humans, compared to the current 3%. He outlined three foundational pillars for the machine intelligence era: AI models, AI chips, and AI cloud. Alibaba's chip subsidiary Pingtouge released the Zhenwu V900 AI chip with triple the computing power of its predecessor, and Alibaba Cloud set a target to operate over 20GW of global data center capacity by 2032. The conference emphasized that AI is now being deployed in specific industrial roles, from factory production lines to legal contract review, rather than just general-purpose tasks.
Read sourceAlibaba Unveils AI Chip, Cloud Plans at 2026 Apsara Conference
At the 2026 Apsara (Yunqi) Conference in Hangzhou, Alibaba Group showcased its full-stack AI strategy, from chips to cloud services. CEO Wu Yongming stated that customer AI demand is strong and that Alibaba will invest heavily in AI infrastructure, targeting over 20GW of global data center capacity by 2032. Chairman Joe Tsai emphasized the importance of scaling AI from breakthroughs to real-world value. Key announcements included the T-Head Zhenwu V900 AI chip, with 3x the computing power of its predecessor, and a roadmap for future server CPUs. Alibaba Cloud CTO Li Feifei detailed the Agentic Cloud strategy, launching new platforms like AgentCore and Agent Sandbox. Wu Yongming predicted that machine thinking will eventually exceed human thinking by 1,000 times, comparing the current AI era to the early days of electricity. The Qwen team is working on models up to 5-10 trillion parameters, aiming towards superintelligence.
Read sourceShow 17 older updatesHide older updates
Liu Dayiheng's Apsara Debut and Alibaba's Agent Bet on Cloud Computing
The 2024 Apsara Conference, Alibaba Cloud's annual event, showcased a strategic shift from selling computing resources to selling task-completion via AI agents. Alibaba CEO Wu Yongming stated that machine thinking will eventually exceed human thinking by 1,000 times, comparing the current AI era to the early days of electricity. The conference featured the first public display of the Panjiu AL128 super server with 128 self-developed Zhenwu M890 chips, and the debut of Liu Dayiheng as Qwen large model project lead. Key demonstrations included an agent controlling a waste-to-energy plant in Wuhan, achieving over 98% automated operation, and an in-car agent coordinating cross-platform tasks. Alibaba unveiled a five-layer full-stack architecture (chip, cloud infrastructure, Qwen model, model services, agentic applications) designed for agent runtime. The company has committed over 380 billion yuan ($52.5 billion) in AI infrastructure over three years, targeting $100 billion in annual cloud and AI revenue within five years. The article notes that cloud computing's pricing unit is shifting from machine time to task completion.
Read sourceAlibaba CEO outlines AI economics strategy with focus on chips, models, and cloud infrastructure
At the 2024 Hangzhou Yunqi Conference, Alibaba Group CEO Wu Yongming delivered a keynote speech shifting focus from AI technology to its economic implications. He argued that machine-generated thinking will eventually exceed human thinking by 1,000 times, drawing a parallel to how industrial revolution commoditized physical power. Wu identified three foundational pillars for the 'machine intelligence era': AI models, AI chips, and AI cloud, which together form a production chain for 'thinking as a commodity.' The article interprets this as Alibaba's strategic bet on becoming the economic infrastructure for the AI age, with chip production (via T-Head), model quality (via Qwen), and cloud distribution (via Alibaba Cloud) as key components. Wu also noted that current AI applications like coding are analogous to early electric lights—proving utility but not yet defining a new era. The analysis concludes that Alibaba is pursuing a dual-track strategy: long-term investment in foundational AI infrastructure and near-term optimization for enterprise agent deployment.
Alibaba Pledges Heavy Investment in AI Models, Chips, and Cloud at 2026 Yunqi Conference
At the 2026 Yunqi Conference on September 22, Alibaba Group CEO Wu Yongming announced the company's strategic commitment to building three foundational pillars for the machine intelligence era: AI models, AI chips, and AI cloud. Wu stated that machines are becoming the primary thinking force and that intelligence is becoming a scalable commodity, with total machine thought expected to exceed human thought by 1,000 times. He emphasized that sufficient AI infrastructure is a prerequisite for this. Alibaba's Qwen team is exploring recursive self-improvement (RSI) technology, where models identify their own weaknesses and evolve through self-designed experiments. The company also released the Zhenwu V900 AI chip, developed by its T-Head semiconductor unit, offering three times the computing power of its predecessor and supporting clusters of up to 500,000 cards. Wu noted strong customer AI demand driving Alibaba Cloud revenue growth, but said global supply chain shortages for AI data centers limit computing power expansion. Alibaba plans to work with partners to build a global AI cloud network.
Read sourceAlibaba CEO Wu Yongming: Future machine thinking will exceed human total by 1,000 times
At the 2026 Hangzhou Yunqi Conference on September 22, Alibaba Group CEO Wu Yongming delivered a speech on the 'machine intelligence' era. He argued that machines are becoming the primary thinkers and that intelligence is becoming a commoditized good, predicting that the total amount of machine thinking will exceed that of humans by more than 1,000 times. He compared this shift to the Industrial Revolution, which turned power into a commodity. Wu stated that the representative product of the machine intelligence era has not yet emerged, likening current AI coding to the electric light bulb in 1882. He outlined three pillars of the machine intelligence era: AI models, AI chips, and AI cloud. Alibaba announced progress including: the Qwen team's advances in recursive self-improvement (RSI) with plans for 5-10 trillion parameter models; the release of the Zhenwu V900 AI chip by T-Head, with 3x performance over the M890 and scalability to 500,000 cards; and a target for Alibaba Cloud to operate over 20 GW of global data center capacity by 2032, citing strong customer AI demand.
Read sourceAlibaba accelerates AI ecosystem integration, unveils chip and model roadmap at Cloud Summit
At the 2026 Hangzhou Cloud Summit, Alibaba Group CEO Wu Yongming declared the arrival of the 'machine intelligence era,' predicting machine thinking will exceed human thinking by 1,000 times. He identified AI models, AI chips, and AI cloud as the three pillars of this era, with Alibaba committed to long-term investment in these areas. Alibaba's semiconductor arm Pingtouge unveiled the Zhenwu V900 AI chip, claiming 3x the compute power of its predecessor, with mass production planned for Q1 2027. The company also revealed a roadmap for its Yitian server CPUs, including the Yitian 720 and 730 in 2027. The Qwen team announced progress on recursive self-improvement (RSI) and plans to scale models to 5-10 trillion parameters. Alibaba Cloud aims to operate over 20GW of global data center capacity by 2032. Analyst Xie Siyuan from Yijing Capital noted that AI industry concentration is increasing among giants like Alibaba and ByteDance, who can integrate the entire AI supply chain. Downstream applications, such as Honor's Magic 9 smartphone integrating Qwen Intelligence, show early commercialization progress.
Read sourceAlibaba CEO Wu Yongming Bets on Machine Intelligence, Says Machine Thinking Will Reach 1,000 Times Human Levels
At the 2026 Hangzhou Yunqi Conference on September 22, Alibaba Group CEO Wu Yongming made a bold forecast that total machine thinking will eventually exceed human thinking by 1,000 times, compared to less than 3% today. He compared the current AI era to the early days of electricity, arguing that today's AI applications like coding and report writing are merely replacing existing work, not creating a new era. Wu outlined Alibaba's strategy to build foundational infrastructure first, including AI models, AI chips, and AI cloud services. The company plans to train a new model with 5-10 trillion parameters, released a new AI chip called Zhenwu V900 with 3x performance over its predecessor, and aims to operate over 20GW of global data center capacity by 2032. Wu emphasized that chip, model, and cloud development are now converging into an integrated system to optimize task performance, response speed, throughput, and cost. The article positions Alibaba's full-stack approach as unique in China, contrasting with Huawei and Baidu's different strategic focuses.
Read sourceAlibaba CEO Wu Yongming: Machine thinking will reach 1,000 times human levels, bets big on AI infrastructure
At the 2026 Hangzhou Yunqi Conference on September 22, Alibaba Group CEO Wu Yongming made a bold prediction: the total amount of machine thinking will eventually reach more than 1,000 times that of humans, compared to less than 3% today. He compared the current state of AI to the early days of electricity, arguing that today's AI applications like coding and writing reports are merely replacing existing work, not creating a new era. To prepare for this future, Alibaba is aggressively building infrastructure, treating AI models, AI chips, and AI cloud as the three pillars of the machine intelligence era. The company announced plans for its Qwen team to train models with 5-10 trillion parameters, and for Alibaba Cloud to operate over 20GW of global data center capacity by 2032. Alibaba's chip unit, Pingtouge, released the new Zhenwu V900 AI chip, claiming 3x the performance of its predecessor. Wu emphasized that the real AI era has not yet arrived, and that Alibaba's full-stack strategy—from chips to models to cloud—is designed to create a synergistic 'flywheel' for future machine intelligence demands.
Read sourceAlibaba Unveils Full AI Stack at Yunqi Conference, Reveals Chip and Cloud Roadmap to 2032
At the 2026 Yunqi Conference in Hangzhou, Alibaba showcased its full-stack AI capabilities spanning chips, cloud, models, and agents. CEO Wu Yongming stated that customer AI demand is strong and that Alibaba will invest heavily in AI infrastructure, targeting 20GW of global data center capacity by 2032. Chairman Joe Tsai emphasized the importance of real-world AI deployment and value creation. Alibaba's chip unit Pingtouge announced the Zhenwu V900 AI chip with 3x the compute of its predecessor, scalable to 500,000 cards, and plans for the Yitian 720 and 730 server CPUs by 2027. CTO Li Feifei detailed the Agentic Cloud strategy with new products like AgentCore and Agent Sandbox. Wu Yongming predicted that machine thinking will exceed human thinking by 1,000 times, comparing the current AI era to the early days of electricity. He outlined three pillars for the machine intelligence era: AI models, AI chips, and AI cloud. The Qwen team is working on recursive self-improvement toward models with 5-10 trillion parameters, aiming for superintelligence.
Read sourceAlibaba Pushes AI as Public Utility at 2026 Cloud Conference, Reveals Full-Stack Ambitions
At the 2026 Hangzhou Cloud Computing Conference, Alibaba Group CEO Eddie Wu unveiled a vision of AI 'thinking' becoming a commodity like electricity, with machine thinking projected to exceed human levels by 1000 times. Wu outlined Alibaba's strategy centered on three pillars: AI models, AI chips, and AI cloud, with a long-term goal of operating over 20GW of data center capacity by 2032. The conference showcased the Qwen3.8-Max language model, the Zhenwu V900 AI training-inference chip (2027 production), and the Agentic Cloud platform. Alibaba positioned itself as China's only full-stack AI company with vertical integration from chips to software. The article critically assesses the claims, noting that many 'first' rankings come from disparate benchmarks, the recursive self-improvement (RSI) narrative remains unverified, and supply chain constraints pose significant risks to the ambitious 20GW and 500,000-chip scaling targets. Enterprise adoption was highlighted through partnerships with Yihai Kerry, FAW Toyota, and others, alongside new hardware like the QwenNote A2 recorder.
Read sourceAlibaba CEO Wu Yongming: Today's AI Is Like the Electric Light in 1882
At the 2026 Hangzhou Yunqi Conference, Alibaba Group CEO Wu Yongming delivered a speech arguing that the term 'artificial intelligence' is inadequate for describing current and future machine intelligence. He stated that machine intelligence is not a simple substitute for human intelligence but a different species, with only about 3% of its potential currently developed. Wu predicted that the total amount of thinking supplied by machines will eventually exceed human thinking by more than 1,000 times. He compared the current state of AI to the electric light in 1882, noting that the representative products of the machine intelligence era have not yet emerged. Wu identified three foundational pillars for the machine intelligence era: AI models, AI chips, and AI cloud. He revealed that Alibaba's Qwen team is exploring recursive self-improvement (RSI) toward artificial superintelligence (ASI) and plans to train models with 5 to 10 trillion parameters. Alibaba's chip unit, Pingtouge, released the 'Zhenwu V900' AI chip, claimed to be the most powerful domestically developed AI chip, set for mass production in Q1 2027. Wu also announced a target for Alibaba Cloud to operate over 20 GW of global data center capacity by 2032.
Read sourceAlibaba Cloud Conference Highlights AI Self-Improvement, Chip Breakthroughs, and New Devices
The 2026 云栖大会 (Cloud Computing Conference) opened in Hangzhou on September 22, themed 'Wisdom for Application,' focusing on Agentic AI across chips, cloud infrastructure, models, and applications. Alibaba Group CEO Wu Yongming outlined a three-pillar AI infrastructure strategy, predicting machine thinking will exceed human thinking by 1,000 times. Alibaba's Qwen team reported progress in Recursive Self-Improvement (RSI), with Qwen3.8-Max self-iterating for over a month without human involvement, improving its score from 40 to 45. Future models Qwen4.5 and Qwen5 are planned with 5-10 trillion parameters. Chip subsidiary Pingtouge unveiled the Zhenwu V900 AI chip, offering 3x the performance of its predecessor, with a single cluster expandable to 500,000 cards, mass production slated for Q1 2027. IDC forecasts China's intelligent computing power will grow 87.9% in 2026, while global AI compute supply-demand gap may widen to $380.9 billion by 2030. AI terminals were a highlight: Honor and Alibaba announced joint development of five vertical models for next-gen AI phones; Alibaba's Qwen Book AI-native PC and Qwen AI glasses N1 series were showcased, with the glasses launching October 13. Zhipu AI disclosed its first engineering RSI case, and OpenAI called for global safety standards on RSI.
Read sourceYunqi Conference 2026: Compute Gap Accelerates Industry Chain, Alibaba Unveils AI Chip and RSI Progress
The 2026 Yunqi Conference opened in Hangzhou on September 22, themed 'Intelligence for Application,' focusing on Agentic AI across chips, cloud infrastructure, models, and applications. Alibaba Group CEO Wu Yongming outlined a three-pillar AI strategy (models, chips, cloud), predicting machine thinking will exceed human thinking by 1,000 times. Alibaba's Qwen team reported progress in recursive self-improvement (RSI), with Qwen3.8-Max iterating autonomously for over a month. Chipmaker Pingtouge unveiled the Zhenwu V900 AI chip, 3x the performance of its predecessor, with mass production slated for Q1 2027. IDC forecasts China's AI computing power will grow 87.9% in 2026, while the global AI compute supply-demand gap may widen to $380.9 billion by 2030. AI terminals were a highlight, including the Qwen AI glasses N1 series and the Qwen Book AI-native PC. Honor and Alibaba announced a partnership on five vertical models for next-gen AI phones. Open AI called for global safety standards on RSI.
Read sourceAlibaba CEO Wu Yongming: Humanity Will Enter the 'Machine Intelligence' Era
At the 2026 Hangzhou Yunqi Conference on September 22, Alibaba Group CEO Wu Yongming delivered a speech outlining his vision for a 'machine intelligence' era. He argued that AI is evolving faster than anticipated, with autonomous improvement paths becoming clear, and that machines are becoming the primary source of thinking, turning intelligence into a scalable commodity akin to power during the Industrial Revolution. Wu made two key predictions: first, that the total amount of machine-generated thinking will exceed human thinking by a factor of 1,000 or more, eventually handling 99.9% of cognitive work; second, that the defining product of the machine intelligence era has not yet emerged, comparing current AI coding to the early electric light bulb in 1882. He stated that the era's foundations are AI models, AI chips, and AI cloud infrastructure, which Alibaba is actively developing. Alibaba Chairman Joe Tsai added that AI is shifting from chasing technical benchmarks to large-scale real-world deployment.
Read sourceAlibaba CEO Wu Yongming Outlines AI Strategy: Thinking as a Commodity Like Electricity
At the 2024 Hangzhou Yunqi Conference, Alibaba Group CEO Wu Yongming delivered a keynote speech framing the future of AI through an economic lens, coining the concept of 'thinking as a commodity'. He argued that machine-generated thinking will eventually exceed human thinking by a factor of 1,000, similar to how machine power replaced physical labor. Wu identified three foundational pillars for this future: AI models, AI chips, and AI cloud infrastructure, which together form a production chain for 'thinking'. The article analyzes Alibaba's strategic bets, including its chip development (e.g., TrueCore V900), the Qwen model family targeting 5-10 trillion parameters, and a plan to operate over 20GW of global data center capacity by 2032. It also details Alibaba's full-stack optimization for enterprise AI agents, addressing compute efficiency, long-task reliability, and cost. The author notes that Wu views current AI applications like AI coding as analogous to the early light bulb (1882), proving the technology's utility but not yet defining a new era. Alibaba's strategy is to build the infrastructure for a future 'killer app' that has not yet emerged.
Read sourceAlibaba AI Strategy: Heavy Compute Investment, Model Leadership, Flexible Application Exploration
Over the past six months, Alibaba's AI business has undergone rapid change, with AI and cloud becoming key growth drivers. Alibaba Cloud reported FY2026 revenue of 158.1 billion yuan (up 33.9% YoY) and Q1 FY2027 AI cloud revenue of 48.4 billion yuan (up 44.9% YoY). CEO Wu Yongming outlined a long-term strategy of heavy investment in AI models, AI chips, and AI cloud infrastructure, targeting 20GW of global data center compute capacity by 2032. Alibaba has committed 380 billion yuan over three years for AI infrastructure, with 190 billion yuan spent by June 2026. The company is also advancing its self-developed Zhenwu V900 AI chip, set for mass production in Q1 2027. On the model side, Alibaba's Qwen series maintains a leading position in the developer community, with over 30 billion downloads and 460 open-source models. The article notes that Alibaba is using a 'bottom-up certainty, top-layer exploration' approach: heavy investment in compute and models provides a stable foundation, while application-layer products (e.g., Qianwen Office, Qoder) are allowed to iterate and compete. The company faces internal organizational challenges and market competition from DeepSeek, Zhipu AI, and others, but believes that securing compute supply and model leadership are the most certain bets in the current AI landscape.
Read sourceAlibaba CEO Wu Yongming: Machine Intelligence Era Will Make Thinking a Scalable Commodity
At the 2026 Hangzhou Yunqi Conference, Alibaba Group CEO Wu Yongming delivered a speech outlining his vision for the 'machine intelligence' era. He argued that machines are becoming the primary agents of thinking, and that 'thinking' will become a scalable commodity, similar to how 'power' became scalable during the Industrial Revolution. Wu forecast that the total amount of machine thinking will be 1,000 times that of human thinking in the future, with current machine thinking at less than 3% of human levels, implying tens of thousands of times growth potential. He stated that the prerequisite for scalable machine thinking is sufficiently developed AI infrastructure, identifying AI models, AI chips, and AI cloud as the three cornerstones of the machine intelligence era. Alibaba will firmly invest in these three areas as a long-term strategic choice. Wu announced progress on the Qwen model team's recursive self-improvement (RSI), plans to expand models to 5-10 trillion parameters, and a comprehensive layout for data center chips including the 'Zhenwu' GPU, 'Yitian' CPU, 'Panmai' smart NIC, and ICN interconnect chips. He also set a target for Alibaba Cloud to operate over 20 GW of global data center capacity by 2032 to meet surging AI demand.
Read sourceAlibaba CEO Wu Yongming: Machine intelligence era requires AI model, chip, cloud investment
At the 2026 Hangzhou Yunqi Conference on September 22, Alibaba Group CEO Wu Yongming declared that humanity is entering a 'machine intelligence' era where thinking becomes a scalable commodity, akin to power in the Industrial Revolution. He forecast that total machine thinking will reach 1,000 times that of humans, with current levels below 3% implying tens of thousands of times growth potential. Wu identified three foundational pillars: AI models, AI chips, and AI cloud. On models, Alibaba's Qwen team is exploring recursive self-improvement (RSI) toward superintelligence (ASI) and plans a 5-10 trillion parameter model. On chips, Alibaba's Pingtouge released the Zhenwu V900 GPU, claiming 3x the compute of its predecessor, scalable to 500,000 cards per cluster. On cloud, Alibaba aims to operate over 20 GW of global data center capacity by 2032, with current AI demand far outstripping supply. Wu emphasized joint optimization of chips, servers, networks, models, and inference systems to boost token throughput.
Read source