English translation is being prepared.
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
The 11th edition of the e-Conomy SEA report, released by Google, Temasek and Bain & Company, forecasts that Southeast Asia's digital economy GMV will grow 17% year-on-year to $368 billion in 2026 and surpass $1 trillion by 2035. Generative AI is already used daily by 243 million people in Southeast Asia, growing four times faster than the mobile internet, and the report projects that regional AI usage will grow 60-fold over the next five years. 75% of users are willing to let AI agents compare prices and complete transactions on their behalf, but the region will need to expand data center capacity roughly threefold over the next five years to support computing demand.

The 11th edition of the e-Conomy SEA report, released by Google, Temasek and Bain & Company, forecasts that Southeast Asia's digital economy GMV will grow 17% year-on-year to $368 billion in 2026 and surpass $1 trillion by 2035. Generative AI is already used daily by 243 million people in Southeast Asia, growing four times faster than the mobile internet, and the report projects that regional AI usage will grow 60-fold over the next five years. 75% of users are willing to let AI agents compare p
Matched fieldsBody
Matched fieldsBody
Amulya Vishwanath, founder and managing partner of Techable Ventures, believes that physical AI is still in its "LEGO era" and has yet to see its ChatGPT moment in robotics. She points out that foundational software and components such as robotic hands and actuators are being redesigned, and that general-purpose and specialized robots are advancing in parallel; these building-block components are key to scaling innovation. She also emphasizes that talent is the core barrier in the U.S. market, and says her network built up over many years can help founders connect with top talent.
Amulya Vishwanath, founder and managing partner of Techable Ventures, believes that physical AI is still in its "LEGO era" and has yet to see its ChatGPT moment in robotics. She points out that foundational software and components such as robotic hands and actuators are being redesigned, and that general-purpose and specialized robots are advancing in parallel; these building-block components are key to scaling innovation. She also emphasizes that talent is the core barrier in the U.S. market, a
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
After its stock price rose sharply for consecutive sessions, Unitree Robotics issued a risk warning, stating that operations are normal but prudently reminding the market about the commercialization progress of humanoid robots. From January to September 2025, its humanoid robot revenue was approximately 595 million yuan, accounting for 51.53% of total revenue, with research and education contributing over 70%; during the same period, gross margin fell from 68.44% in 2024 to 62.91%. As of October 20, Unitree's trailing price-to-earnings ratio stood at 388.25 times, far above the general equipment manufacturing industry average of 39.51 times.
After its stock price rose sharply for consecutive sessions, Unitree Robotics issued a risk warning, stating that operations are normal but prudently reminding the market about the commercialization progress of humanoid robots. From January to September 2025, its humanoid robot revenue was approximately 595 million yuan, accounting for 51.53% of total revenue, with research and education contributing over 70%; during the same period, gross margin fell from 68.44% in 2024 to 62.91%. As of October
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Matched fieldsBody
Memory Tensor has open-sourced and released the MemT 1.0 series of memory-native models, including MemT-1.0-Nano with 36B parameters and 3B activated (based on Qwen3.6-35B-A3B) and MemT-1.0-Flash with 288B parameters and 17B activated (based on DeepSeek-V4-Flash), adding a memory computation pathway parallel to attention in selected Transformer modules. The team built a 34-billion-token memory training corpus, of which 20 billion tokens were used for continued pre-training and 14 billion for post-training. In a comprehensive evaluation across 11 memory benchmarks, Flash and Nano achieved average scores of 50.6 and 47.0, respectively, ranking in the top two among the parametric memory methods evaluated; on 500 SWE-bench Verified tasks, the proportion of tasks that Nano succeeded on all three attempts rose from 54.00% for the no-experience baseline to 62.80%.
Memory Tensor has open-sourced and released the MemT 1.0 series of memory-native models, including MemT-1.0-Nano with 36B parameters and 3B activated (based on Qwen3.6-35B-A3B) and MemT-1.0-Flash with 288B parameters and 17B activated (based on DeepSeek-V4-Flash), adding a memory computation pathway parallel to attention in selected Transformer modules. The team built a 34-billion-token memory training corpus, of which 20 billion tokens were used for continued pre-training and 14 billion for pos
Matched fieldsBody