跳到正文
今天10月11日周日26 条
  1. 钛媒体43

    Meitu's roller-coaster ride: AI revival meets rising doubts

    Meitu's stock has fallen more than 50% in 2026, dropping from above HK$9 to around HK$4 after years of AI-driven gains. Its generative AI model MiracleVision helped lift revenue from continuing operations to RMB 2.213 billion in the first half of 2026, up 22.1% year over year, with paid subscribers exceeding 18.44 million. But intensifying competition from ByteDance's CapCut and Doubao, Alibaba's Qwen, and rising compute costs — RMB 134 million in cloud and compute expenses in H1 2026 — are weighing on margins and sentiment.

  2. 钛媒体53

    China AI starts doing the math on compute: data centers, price cuts and Alibaba's capex

    An opinion analysis argues China's AI competition is shifting from acquiring compute to operating it as a capital-intensive business. It cites FT reporting on 89 contracted data center projects in Ulanqab, DeepSeek's permanent 75% API price cut for V4-Pro announced in May, a Silicon Data index showing token prices down over 52% from May highs, and Alibaba's quarterly results of RMB 67.678 billion capex up 75% year over year alongside RMB 48.437 billion AI cloud and compute revenue up 45%. The piece notes model prices can adjust in months while data center investments take years to recoup.

  3. 钛媒体44

    Golden Week OTAs vs AI assistants: Tongcheng and Fliggy sell inventory while Doubao and Qwen sell answers

    During China's National Day Golden Week, AI travel entry points became a battleground between traditional OTAs and AI assistants. Doubao launched a dedicated travel hub (“出行用豆包”) with four modules—maps, transport, hotels, and dining/entertainment—and hotel bookings via Doubao rose over 56% week-over-week; Fliggy's AI assistant saw interactions up 158% and flight-hotel bookings up 110%, while Meituan's AI管家 “小团” searches rose 29% and Tongcheng's DeepTrip agent saw travel queries jump 50%+ in the first four days. The piece frames this as “LLM+” versus “+LLM” strategies, arguing the decisive factor will be fulfillment and supply-chain capability rather than conversational entry points.

  4. 钛媒体38

    Care robots for the elderly: how far are we from machines that can care for aging humans?

    Robots may become the realistic answer to elder care as populations age and caregiving labor grows unsustainable—Japan's health ministry recorded at least 486 deaths of people over 65 from caregiver abuse between 2006 and 2024. Today's robots handle monitoring, reminders, conversation, and companionship well, but physical labor like lifting or bathing the elderly remains the hardest frontier due to scarce real-world contact data; Japan's two decades of nursing-robot policy show specialized devices like Cyberdyne's HAL exoskeleton survive while general-purpose humanoids like RIBA and Robear stay prototypes. The likely path is institutional adoption first and homes later, with liability and payment constraints shaping the rollout.

10月10日周六
  1. 钛媒体48

    AI Ride-Hailing: How AI Agents Are Reshaping China's Mobility Industry Chain

    AI-driven ride-hailing is taking shape in China, with Cao Cao Mobility in September integrating its AI booking service into system-level agents including the Doubao phone assistant on the Nubia NaviX Ultra, Honor's YOYO, OPPO's Xiao Bu Next, and Qwen AI glasses, enabling full booking without opening any app. China's ride-hailing market reached roughly RMB 411.9 billion in 2025, up 6.07% year over year, with users flat at 539 million, pushing platforms toward AI agents as a new growth story. Meanwhile, Robotaxi operators like Apollo Go (over 23 million cumulative orders in 28 cities), Pony.ai, and WeRide are moving toward mixed operations and eventual integration with ride-hailing platforms.

板块