Unitree Technology, widely described in the source article as China’s "first humanoid robotics stock," drew intense demand in its IPO, with an online allotment rate of just 0.01809759% after a callback mechanism and an oversubscription multiple of 8,288.82 times. The company priced its shares at RMB 150.8, issued 40.4464 million shares, and raised about RMB 6.099 billion, exceeding its original fundraising target by 45.15%. Pre-listing perpetual contract data from trade.xyz implied a market capitalization of about $35.47 billion, or roughly RMB 239 billion, while predict.fun showed a 75% probability that Unitree’s closing market value would top RMB 200 billion. The article also lays out a wide range of post-listing valuation views, from RMB 109 billion to more than RMB 400 billion, and highlights founder Wang Xingxing’s responses on valuation, shipment scale, embodied AI capabilities, and criticism over remote control features.

Coinbase is using an internal AI coding platform called Forge to explore a different model for software development, one built around multiple AI agents working in parallel rather than a traditional step-by-step workflow. The system has evolved from an early internal tool into a broader ecosystem of AI development capabilities and is now described as an important part of the open-source AI coding framework OpenSWE. According to the information disclosed, Forge was previously known as Cloudbot, and an even earlier version was called Claudebot. OpenSWE is designed to provide infrastructure resembling an “AI software engineer,” with agent orchestration, cloud sandboxes, tool use, integrations with Slack, Linear, and GitHub, support for sub-agents, and automatic Pull Request creation. One of Forge’s most watched features is Mux, which has about 600 internal users at Coinbase and lets engineers run multiple AI agents at the same time. Coinbase engineering lead Chintan Turakhia also ran an internal experiment in which nearly 1,000 engineers were asked to stop using traditional IDE tools for two weeks to see what parts of development an AI-driven workflow could automate. He said Coinbase’s AI spending has fallen as use of coding tools has increased, while token usage has continued to rise.