Summary 82 | Bouygues Asia

Newsletter N°82 - November 2025

📉 Economic Outlook: China’s 2025–2029 Plan: Xi Jinping Consolidates Control and Sets New Priorities

At the end of October, the 4th Plenum of the 21st Central Committee of the CCP was held to prepare the 2025–2029 plan (covering economic, social, ideological, and environmental trajectories), which will be formally adopted early next year. It was also an occasion to display Xi Jinping’s firm grip on the committee, as 37 seats out of 205 were prominently shown empty on TV, likely recently purged, marking the largest purge since the Mao era.

Xinhua News Agency published the full text of the draft five-year plan immediately after the end of the plenum. An explanation written by Xi Jinping himself was also quickly released. Interestingly, no firm figures for GDP growth were mentioned, as trade frictions with the US create a high level of uncertainty. Xi only reiterated the long-term goal of bringing China’s per-capita GDP “on a par with a mid-level developed country” (likely referring to countries such as Spain or Italy, whose per-capita GDP stands around USD 35,000–40,000; for reference, China’s is currently closer to USD 13,500).

Among the remarks that could shape the overall orientation of the financial and economic measures to come, Xi stated that the country must “address rat-race competition through holistic measures,” a likely reference to the intensifying price competition caused by shrinking domestic demand and resulting oversupply. Furthermore, in the financial sector, the plan states that China will continue to develop the “digital renminbi” as a centralized digital central-bank currency.

As strategic priority sectors for the upcoming five years, the plan highlights the strengthening of the semiconductor and AI industries. The plenum also designated the space sector as a top priority for the first time, pledging to turn China into a leading space nation within the next five years.

Measures to ease the demographic crisis were also mentioned (including the reduction of educational burdens on families and the expansion of free education). Last but not least, the goals of reaching peak carbon emissions and progressing toward carbon neutrality were reiterated.

🏗️ Smart Construction: Samsung C&T Showcases Five Construction Robots at Raemian Robot Week 2025

Samsung C&T’s Engineering & Construction Group held a demonstration event showcasing its latest housing construction robots, highlighting the company’s commitment to innovation and digital transformation in the construction industry.

On October 30, Samsung C&T hosted “Raemian Robot Week 2025” at the Banpo 3 Redevelopment Site in Seoul, unveiling five types of construction robots designed to improve efficiency and safety on housing sites.

Five Construction Robots for Safer and Smarter Housing Sites

The five robots demonstrated were:

Autonomous forklift
Material transport robot
Cleaning robot
Water-spraying drone
Wearable robotic exoskeleton

These technologies are expected to significantly reduce dangerous manual tasks, streamline logistics, and improve productivity at housing construction sites.

Autonomous Material Handling and Nighttime Operations

The autonomous forklift and material transport robot, developed by Seoul Dynamics, operate automatically during nighttime hours when no workers are on-site. This minimizes interference with daytime construction processes while improving operational efficiency.

The autonomous forklift identifies palletized materials stored on-site and delivers them to designated drop-off points within each apartment building. The material transport robot then loads the materials, calls the elevator, rides it autonomously, and unloads supplies directly inside individual apartment units.

To enable this, Korea Mitsubishi Elevator has installed a communication module at the Banpo 3 Redevelopment site that allows elevators and robots to interact part of an ongoing field test to validate robot integration in construction environments.

Drone and Cleaning Robots for Safer, Cleaner Environments

These technologies are expected to significantly reduce dangerous manual tasks, streamline logistics, and improve productivity at housing construction sites.

Autonomous Material Handling and Nighttime Operations

Drone and Cleaning Robots for Safer, Cleaner Environments

Wearable Robot to Support Construction Workers

Building a Collaborative Construction Robotics EcosystemSamsung C&T emphasized that the event served as a test-bed collaboration with domestic robotics firms possessing advanced technologies. By opening real construction sites for testing and demonstration, the company enables developers to refine and optimize robots for actual field conditions.

Kim Myung-Seok, Executive Vice President and Head of the Housing Division at Samsung C&T, stated:

“This demonstration event is part of our broader effort to establish a comprehensive housing construction robotics ecosystem. We will continue to collaborate with leading robotics companies to develop robots specialized for construction sites, improve productivity, and eliminate safety risks for our workers.”

Through initiatives like Raemian Robot Week 2025, Samsung C&T continues to drive technological innovation in the construction sector reinforcing its vision of a safer, smarter, and more sustainable job site powered by robotics and AI.

🏗️Construction: BA Generative AI System for Construction Site Risk Prediction

Haseko Corporation has developed and put into operation Kizuki AI, a generative artificial intelligence tool designed to assist site managers in anticipating potential hazards. Built on the open-source AI development platform “Dify,” the System aims to standardize knowledge related to on-site safety measures and optimize preventive interventions.

As of August 2025, the System was deployed across all 129 of the Company’s construction sites in the Tokyo metropolitan area, with plans underway to gradually expand its use to all project sites nationwide.

To eliminate construction site accidents, it is crucial that every site manager remains aware of potential risks and takes preventive measures. However, risk assessment often depends on individual experience, leading to possible oversight of certain hazards. Moreover, under current industry challenges such as labor shortages and restrictions on overtime work, there is an increasing need for more effective and data-driven methods to enhance safety practices.

By simply entering details such as the type and nature of the construction work, weather conditions, and temperature on a computer or smartphone, and uploading photos of the site, users can receive immediate alerts from the AI.

The System then generates predictions of possible risks, such as heatstroke or unstable scaffolding—and provides corresponding safety measures, raising awareness among site managers.

When a site manager uses the system, they input project details, weather, and temperature data on the input screen and upload a site photo. The AI’s predicted risks and recommended countermeasures are then displayed immediately on the results screen alongside the image.

🧠 Generative AI: ByteDance's Seed3D 1.0: A Generative AI Leap for Simulation

ByteDance has entered the 3D generative AI arena with Seed3D 1.0, a model designed to create simulation-ready 3D assets from simple 2D images. This technology promises to significantly accelerate development in fields like robotics, manufacturing, and digital twin simulations by automating the often laborious process of 3D asset creation.

Seed 3D Process

The core innovation of Seed3D is its ability to transform a flat image into a detailed, realistic 3D object. These assets are not just visual; they are physics-ready and directly compatible with industry-standard platforms like NVIDIA’s Isaac Sim.

The model uses vision-language AI to ensure objects are scaled to real-world proportions, a critical feature for accurate robotics training and industrial design. Benchmark tests show it outperforms existing models in generating high-fidelity geometry and textures that remain stable during simulated manipulation.

Beyond single objects, Seed3D can assemble entire coherent scenes, analyzing spatial relationships in an input image to arrange multiple assets into a ready-to-use virtual environment. This capability is a major boon for systems integrators and robotics firms, who could use it to build end-to-end pipelines that quickly turn product photos into dynamic, simulation-ready worlds

🚗Mobility: Japan Mobility Show 2025: A Record-Breaking Showcase of Future Mobility Innovation

Japan Mobility Show 2025 opened October 29 at Tokyo Big Sight with unprecedented scale and ambition, attracting over 500 companies and organizations to become the largest automotive event in Japanese history. Running through November 9, the show expands beyond automobiles to encompass IT, telecommunications, and electronics industries, with more than 160 companies participating in the "Tokyo Future Tour 2035" program that visualizes mobility in 2035.

The show's theme—"Find the exciting future!"—emphasizes three pillars: FUTURE (mobility innovation), CULTURE (automotive heritage with 30 vintage vehicles), and PEOPLE (human-centered design), creating an experience for "everyone from children to adults".

Major innovative projects include:

Sharp's LDK+ – Redefining EV Purpose

Sharp's LDK+ (second generation concept) represents the boldest challenge to automotive convention, shifting focus from the car's 5% driving time to its 95% stationary parking hours. This compact minivan-based EV, built on Foxconn's "Model A" platform, features a revolutionary rotating driver's seat that converts the cabin into a living room.

Sharp's LDK+ compact minivan with integrated projector technology

The second-generation refinement downsizes from the first model's seven-seater to a more agile compact minivan while maintaining spacious interiors.

Interior amenities include:

Integrated projector and roll-out screen for theater and remote work applications
AI home control integration via Sharp's "COCORO HOME" platform, connecting kitchen appliances, air conditioning, and washing machines to create a seamless living space
Flexible configurations from solitude spaces for teenagers to cozy corners for seniors

BYD's "Living Car" – Chinese EV Ecosystem ExpansionBYD's commercial vehicle showcase includes the revolutionary "J6 Living Car"—a truck reimagined as a mobile living and working space. This marks China's most ambitious Japanese market penetration yet, with the company exhibiting 13 vehicles including the world-first "RACCO" lightweight EV and the premium "YANGWANG U9" supercar.

The living car concept—complete with sleeping, cooking, and entertainment facilities—targets Japan's aging driver population and emerging "van life" culture. BYD's strategy combines aggressive pricing with lifestyle repositioning, positioning trucks as not merely commercial tools but integrated living solutions.

Nissan's Manga-Inspired AI Integration

Nissan's booth design, inspired by Japanese manga aesthetics, represents a paradigm shift in human-machine interfaces. The installation merges monochromatic graphic novel styling with cutting-edge AI and sensor technology, creating an interactive space where visitor engagement "completes the artwork".

This approach signals automotive industry recognition that future vehicles compete on cultural appeal and design philosophy, not purely technical specifications. Nissan's emphasis on "coming alive through visitor interaction" suggests EVs will function as social canvases for expression.

Market Context: Japan's EV Acceleration

The 2025 show coincides with unprecedented domestic EV launches: six new models including Honda's first lightweight EV, Toyota-Daihatsu-Suzuki's joint commercial vehicle, and international entrants from BYD and Hyundai. Japanese manufacturers are simultaneously defending domestic market share while embracing foreign competition—a calculated strategy signaling confidence in EV market maturation.

With over 500 companies committed to mobility transformation, Japan Mobility Show 2025 has established itself as the definitive global benchmark for automotive and mobility innovation.

📺TV: Samsung Turns TVs into AI Companions with Perplexity Integration

Samsung Electronics Co. announced that it will become the first company in the industry to integrate the AI-powered search engine Perplexity into its 2025 lineup of TVs and monitors, marking a major milestone in transforming traditional displays into intelligent, interactive AI screens.

This integration represents a significant evolution in the TV and monitor experience moving beyond simple voice command recognition to devices capable of understanding natural language, generating contextual answers, and suggesting follow-up questions.

A New Kind of AI-Powered Screen

Perplexity is an AI-based generative search engine that collects and analyzes information from reliable sources to deliver in-depth, conversational answers. Unlike traditional search boxes, it operates in a dialogue-based format, allowing users to interact with their devices naturally.

On screen: “Ask me anything. To activate the microphone, press …”

For example, users can ask questions such as: “Plan a 4-day trip to Osaka,” “Summarize Samsung Electronics’ third-quarter earnings,” or “Recommend the best movies of 2025.”

Image of the generative AI search results from Perplexity, integrated for the first time in the industry into Samsung Electronics’ 2025 TVs and monitors

Perplexity then provides optimized, well-sourced answers and suggests follow-up questions to deepen the conversation. The AI covers a broad range of topics including movie recommendations, travel planning, sports results, and corporate performance analysis.

Available Models and Access

The Perplexity feature will debut on the 2025 models of:

Micro RGB,
Neo QLED (Quantum-dot Light Emitting Diode),
QLED,
OLED (Organic Light Emitting Diode),
The Frame Pro, and The Frame TV,as well as the M7, M8, and M9 monitors.

Users can easily launch Perplexity by pressing the AI button on the remote control or by opening the “Vision AI Companion” through the Apps tab on their TV or monitor screen.

Customers who purchase Samsung TVs or monitors equipped with Perplexity will also receive a free 12-month subscription to Perplexity Pro, the premium version of the service.

Future Expansion

Samsung Electronics confirmed that Perplexity will not be limited to new models. Through an operating-system (OS) update scheduled for later this year, the company plans to expand support to its 2023 and 2024 TV and monitor lineups, ensuring a consistent AI experience across its ecosystem.

This follows Samsung’s August announcement that Microsoft’s Copilot would also be integrated into the 2025 lineup, underscoring the company’s commitment to enhancing its AI TV experience and transforming displays into intelligent assistants.

At CES 2025 earlier this year, Samsung had already previewed its personalized AI platform “Vision AI Companion,” which aims to expand customized screen experiences through collaborations with global AI leaders, a strategy that now includes its partnership with Perplexity.

Executive Remarks

Ahn Hee-young, Executive Vice President of Samsung Electronics’ Visual Display Business, commented:

“By integrating Perplexity into Samsung TVs and monitors for the first time in the industry, we can now provide a more personalized and exceptional AI screen experience. Through our partnership with Perplexity, we expect users to enjoy their large screens in newer and more meaningful ways.”

Ryan Foutty, Senior Vice President of Business Development at Perplexity, added:

“Our mission is to combine an innovative AI-based interface with traditional search methods to satisfy human curiosity. We’re thrilled to collaborate with Samsung — the world’s No. 1 TV brand — to bring Perplexity to Samsung device users worldwide.”

Toward the Era of the AI Screen

With Perplexity joining Copilot and Vision AI Companion, Samsung is positioning its 2025 product line as the first generation of fully AI-integrated displays devices that don’t just show content, but understand, respond, and assist users in their daily lives.

Through these strategic partnerships, Samsung continues to redefine what a screen can be: a personalized AI companion for information, creativity, and discovery right in the living room.

📡AI: Kakao Integrates OpenAI’s ChatGPT Into KakaoTalk, Delivering Next-Generation AI Functionality to Everyday Messaging

Kakao Corp. announced today the launch of “ChatGPT for Kakao”, a major update that embeds OpenAI’s latest GPT-5 model directly inside KakaoTalk, South Korea’s most widely used messenger app with more than 50 million users.

The integration represents one of the most advanced examples of a global generative-AI engine being natively merged into a local consumer platform, turning KakaoTalk into an intelligent communication hub that merges language understanding, task automation, and multimodal generation within a single chat interface.

Seamless AI Access and Real-Time Responses

With ChatGPT for Kakao, users can now invoke AI assistance instantly inside any chat window—without installing additional apps or logging in separately.

A new ChatGPT icon, positioned at the top of the chat tab, enables users to:

Ask complex or follow-up questions in natural language
Summarize conversations, generate text or images, and translate content
Share AI-generated results directly within chatrooms in real time

The service supports existing OpenAI accounts, as well as new sign-ins through Kakao IDs, with free-tier usage limits and subscription options aligned with OpenAI’s pricing. A launch promotion grants one-month cashback for users subscribing to ChatGPT Plus for three consecutive months through KakaoTalk.

Kakao Tools: The Core Integration Framework

At the technical core of this rollout is Kakao Tools, a modular AI-connection layer that allows ChatGPT to execute contextual actions across Kakao’s ecosystem. At launch, Kakao Tools links ChatGPT to:

Kakao Map — for location searches and navigation
KakaoTalk Reservations — for booking restaurants or events
KakaoTalk Gifts — for direct in-chat gifting
Melon — for music and playlist recommendations

For instance, if a user asks “Find a croffle café near Hapjeong Station,” ChatGPT triggers Kakao Map to display real-time results. A follow-up like “Book a table and share the location with my friends” automatically links to Kakao Reservations, completing the task inside the same chat interface.

Dual-AI Architecture: GPT-5 × Kanana Nano

The integration complements Kakao’s proprietary on-device AI model, Kanana Nano, which powers “Kanana in KakaoTalk” an assistant that understands context, schedules, and proactive suggestions.
Together, GPT-5 and Kanana Nano form a dual-AI architecture balancing cloud-based generative intelligence with local processing for privacy, speed, and personalization.

Future upgrades will extend this architecture to voice commands, multimodal image and video generation, and real-time information retrieval via OpenAI APIs.

Responsible AI and Data Transparency

Kakao emphasized that user trust and transparency remain foundational. Users can control whether their ChatGPT data is stored or used for model training, under Kakao’s and OpenAI’s privacy frameworks that comply with Korean and international standards.

“Our goal is to make AI practical, safe, and deeply integrated into daily life while keeping KakaoTalk’s simplicity and reliability,” a Kakao spokesperson said.

Expanding Beyond Messaging

Kakao plans to expand ChatGPT for Kakao to other group services including Kakao Pay, Kakao Mobility, and Kakao Entertainment and open APIs for third-party developers.
A pilot ‘AI Citizen Secretary’ project with the Ministry of the Interior and Safety is also underway to connect public services through conversational AI.

On the picture: “Meet it directly on Kakaotalk – ChatGPT for Kakao.”

🎬 AI News Media: Japan’s First Copyright-Respecting AI News Search Service

KDDI Corporation (a leading Japanese telecommunications company) in partnership with Google Cloud Japan, will launch an AI-powered article search service in spring 2026 to address the challenges of copyright violations and misinformation often associated with generative AI.

Unlike conventional AI search tools that crawl the entire internet, this service will only use news articles from six participating Japanese media companies—NewsPicks Select (Uzabase), Natalie (Natasha Inc.), Kakaku.com, LDK (Shinyusha), Mansion Note (Renga), and Mamari (Konehito).

Purpose and Motivation
The main objective of this initiative is to tackle copyright violations and misinformation now prevalent in global AI search tools. Unlike typical AI systems that automatically scrape the internet for information, KDDI’s service exclusively indexes articles from a select group of Japanese media companies that join the partnership. Each search result will always display the original source, ensuring that journalists’ and publishers’ rights are respected.

This model responds directly to legal action taken by leading Japanese newspapers including Nikkei, Asahi Shimbun, and Yomiuri Shimbun against US startup Perplexity, which used their articles without permission for AI search functions. The KDDI system aims to prevent such unauthorized use and sets a new ethical standard for AI-powered content discovery.

How the Service Works?

KDDI’s AI, built with Google’s Gemini and NotebookLM technology, will generate summaries and answer queries based only on content from officially sanctioned media partners.
Users can quickly find high-quality, factual reporting, while knowing every piece of information is attributed to its legal owner.
Media companies also benefit directly, earning revenue whenever their content is used in search results, supporting journalism sustainability in the age of generative AI.

🎬 Media & Content: Korean Main Streaming Platform Tving launches New User Interface Focusing on Live Service: From Passive Viewing to Daily Streaming Habit

Strategic Redesign for Habit Formation

Tving announced a comprehensive redesign of its live streaming service on October 30, 2025, positioning itself as a "daily habit OTT" by mimicking social media's addictive design patterns. The overhaul represents a deliberate strategy to combat Korea's booming short-form video consumption, where users spend 48 hours and 73 minutes monthly on short-form platforms compared to just 7 hours and 14 minutes on traditional OTT services.

Revolutionary "Press-and-Play" Interface

The redesigned live service employs a SNS-style auto-scrolling mechanism as its cornerstone feature. When users launch the app, popular live channels auto-play immediately, and vertically scrolling (like TikTok or Instagram) reveals successive popular channels in continuous flow. This contrasts sharply with traditional grid-based channel browsing, creating a passive discovery experience that mirrors TikTok's infamous engagement algorithms.

Key interface innovations include:

Auto-play on launch: Premium content plays instantly without user intervention
Scrollable feed mode: Swipe down to discover additional channels seamlessly
List view toggle: Switch to traditional channel lists for intentional browsing
Program guide integration: Top menu shows scheduled programming for advance planning
Notification system: Set reminders for upcoming content

Tving’s new user interface is all about live service

Massive Content Expansion

Tving expanded its live offering from ~33 channels to approximately 200—a 6-fold increase that positions live streaming as the platform's key differentiator against competitors. The expanded lineup encompasses:

News channels: KBS, YTN, MBN real-time broadcasts
Entertainment originals: tvN, JTBC, OCN, Mnet live programming
Creator content: Independent broadcaster live streams
Sports & events: KBO baseball, various live performances

The "같이볼래?" (Wanna Watch Together?) feature enables synchronized live viewing of dramas and entertainment, while "팬덤중계" (Fandom Broadcast) pairs celebrity hosts with baseball games—positioning sports as social experiences rather than solo consumption.

Personalization Through Behavioral Analytics

Unlike Netflix's algorithmic recommendations, Tving's new system employs viewing pattern analysis to deliver hyper-personalized recommendations combining both live channels and connected VOD content. Users can:

Create custom favorite channel collections
Browse by genre filters (drama, entertainment, animation, news)
Receive AI-curated suggestions based on viewing history
Access related on-demand content from watched live programs

Strategic Context: Korean OTT Consolidation

The redesign comes amid unprecedented consolidation in Korean streaming. The Tving-Wavve merger (approved with pricing restrictions until December 2026) created South Korea's largest combined domestic streaming service, commanding approximately 40% of Korean streaming market share (Tving 16%, post-merger entity targeting 30%+).

Tving's live service overhaul directly addresses research findings that habit formation is critical to OTT retention, with emotional commitment and word-of-mouth intention strongly correlated to habitual daily usage. By engineering "press-and-play" mechanics similar to short-form video platforms, Tving aims to recapture audience attention from YouTube Shorts, TikTok, and Instagram Reels that collectively dominate Korean digital consumption.

Rollout and Future Implications

The redesigned live service launched on Android and PC versions first on October 30, with sequential updates planned for iOS and smart TV platforms. Tving executives emphasized that live streaming represents a crucial competitive moat against global platforms lacking Korea's terrestrial broadcaster partnerships.

This redesign signals a broader industry shift: Korean OTT platforms are competing not on content depth but on engagement mechanics borrowed from social platforms. By converting passive spectators into habitual daily users through SNS-style interfaces, Tving positions itself to capitalize on Korea's sophisticated streaming infrastructure while fending off international competitors investing heavily in Korean market share.