2026-09-11 20:05

Introduction: Model capability leap, compute iteration, and Token economics take center stage. AI is changing. Models are evolving—olympiad gold medals are no longer the ceiling. Within just under ten days at the beginning of September, Anthropic announced the first end-to-end formal proof of Fermat’s Last Theorem in 11 days, followed by OpenAI solving the Millennium Prize Problem—the Navier-Stokes equations—in 88 hours. Meanwhile, AI is becoming more grounded: Chat has shifted from being a "service" to a "service paradigm." Regardless of technical literacy, simply speaking to an AI enables application development and document completion. OpenAI’s newly released GPT-6 Astra even promises full-task automation, direct delivery of results, and continuous maintenance and improvement. More radical experiments are brewing—sources from a model startup revealed that many entrants are already betting on local models and Diffusion models, potentially disrupting the current cloud-based large model dominance.

AI is changing.

Models are evolving—olympiad gold medals are no longer the ceiling. Within just under ten days at the beginning of September, Anthropic announced the first end-to-end formal proof of Fermat’s Last Theorem in 11 days, followed by OpenAI solving the Millennium Prize Problem—the Navier-Stokes equations—in 88 hours. Meanwhile, AI is becoming more grounded: Chat has shifted from being a "service" to a "service paradigm." Regardless of technical literacy, simply speaking to an AI enables application development and document completion. OpenAI’s newly released GPT-6 Astra even promises full-task automation, direct delivery of results, and continuous maintenance and improvement. More radical experiments are brewing—sources from a model startup revealed that many entrants are already betting on local models and Diffusion models, potentially disrupting the current cloud-based large model dominance.

Compute is transforming—AI chips are iterating at unprecedented speed, server rack power densities are rising rapidly, storage and optical modules are triggering market turbulence, and liquid cooling, materials, and manufacturing processes are all undergoing upgrades. A leading liquid cooling firm told us, “chip iteration every year” has triggered systemic competition across the compute stack, with supply chains exploring new co-development models—customers sharing public roadmaps, vendors disclosing technology—compressing three-year cycles into one.

Business models are also shifting. Last year, enterprises chased “Token maximization,” but this year’s “Token economics” emphasizes whether each Token investment yields calculable returns. This has spurred an entire ecosystem of AI full-stack infrastructure services. While emerging verticals bring vitality, they also breed chaos. An AI Infra co-founder lamented the current Token service market as “mixed quality” and “poor reliability”—the same model exhibits performance differences of up to fourfold across different providers in metrics like first-Token latency and throughput. Despite surging demand this year, overall Token service quality in China has noticeably declined. Identifying high-quality Token services in a rapidly shifting market has now become the critical variable determining whether enterprise AI adoption is effective and cost-efficient.

More importantly, our relationship with AI is evolving.

At the individual level, developers and knowledge workers alike must adapt to collaborating—and competing—with Agents; organizationally, “applying AI” is now foundational, while “AI-native” has emerged as the new challenge for organizational evolution; macroscopically, as AI enters the deep waters of scalable deployment, every industry will undergo a new paradigm shift in the pursuit of value realization. Of course, people’s visions of the future are changing too: as embodied intelligence moves toward large-scale commercialization, physical AI is transitioning from fantasy to reality, opening new frontiers of imagination through ongoing technological exploration.

People have always known AI would change—but rarely could they accurately gauge its pace. Discussions about AI deployment began early, yet few were prepared with integration and fusion strategies. Thus, when the AI system transformation sweeps across every corner of business society, some scramble while others seize the advantage.

For enterprises, the key lies in finding their own rhythm of iteration and keeping pace with AI’s rapid evolution.

This requires a deep, authentic understanding of AI and one’s own industry’s frontier and fundamentals, grasping what peers and pioneers are doing and planning to do. It especially demands clarity on three core questions within the technical, industrial, and commercial landscape: Where are we? Where are our partners? Where is the capital?

To answer these questions and stay aligned with rapid changes, the 2026 ITValue Summit Digital Value Summit will continue on September 17, after the main forum concludes, dedicating an entire day to six parallel closed-door sessions. These gatherings will bring top-tier decision-makers from AI and industry together in intimate settings to discuss specific challenges in responding to AI transformation and identify practical, actionable AI implementation strategies.

These closed-door agendas maintain continuity with the main forum’s focus on “capability building,” “ecosystem synergy,” and “scenario integration,” pushing deeper into real-world business operations. CXOs will share tangible insights and judgments, while decision-makers across the value chain will exchange ideas, debate perspectives, and delve into the core pain points, collaboratively crafting viable solutions.

The six closed-door sessions cover every major domain of the AI world: compute, models, Tokens, coding, physical AI, and application deployment. They also drill down vertically from organizational, financial, business, and IT angles, exploring concrete issues and case studies. Over several hours, core players in the industry and major enterprise clients will deliver keynote speeches and showcase benchmark cases, while all participants will engage in open roundtable discussions on the hottest topics in AI implementation.

In addition, a Billion-Dollar Club Thinkers’ Session will focus on the systematic construction and efficacy validation of large-scale AI deployment in enterprises, tackling two core concerns: “How to prove value?” and “How to overcome organizational resistance?” (Note: Guests for this session are invited only.)

Special Closed-Door Session 1 | The Utilities of the Intelligent Era: Token-Powered Full-Stack AI Infrastructure Services

As Tokens become the fundamental unit of measurement in the AI era, the question of how to “buy cheaply, use wisely” has become a critical business imperative.

This closed-door session gathers pioneers in “Token economics,” focusing on how to achieve higher commercial conversion rates and more precise inference outcomes with reduced Token expenditure.

Hot buzzwords in AI implementation will be dissected in depth: Liu Yulong, CTO of FlowerCat Intelligence Studio, will present case studies on Agent applications and the FDE framework for breakthroughs; Jun Wu, Secretary-General of the World Artificial Intelligence Association, will address OPC community building and AI-industry collaborative incubation; Liu Lin, Partner at Opensand, will provide the latest assessment on evolving supply-demand dynamics in model Tokens and compute markets; Chen Zhengqin, Vice President at Method Capital, will share capital insights on the transition from data intelligence to business intelligence; Li Feifan, Partner at Titanium Media Group, will deliver a dedicated analysis on the industry landscape and cross-border market opportunities in AI infrastructure services.

Additionally, the roundtable will explore hot topics such as: “Where to start optimizing model training services?” “In which industrial scenarios does AI Agent deployment yield measurable benefits?” and “What is the current status, challenges, and pain points of FDE and GEO?”

Special Closed-Door Session 2 | CFO × CIO: A New Paradigm for AI Value Realization

CFOs and CIOs are the central figures in this session—and pivotal roles in enterprise governance and organizational transformation. With AI’s support, even without relying on traditional IT tools, CFOs can seamlessly transition from “behind-the-scenes advisors” to “frontline operators,” profoundly impacting internal operations and business execution.

This session directly addresses shifts in power dynamics and organizational logic. CFOs and CIOs will share real-world experiences integrating AI with professional work. For example, Lei Zhirong, Chair of ACCA’s South China Expert Advisory Panel, will present agile development of the commercial insight agent DataAgent; Bu Shengdong, General Manager of Green Town China’s Digital Construction Center, will reveal the keys to enterprise-level AI adoption and value leap.

The roundtable will engage in heated debates around three core themes: “AI project ‘upward management,’” “value realization in AI projects,” and “future organizational restructuring,” delivering firsthand insights into how AI projects begin, progress, and succeed.

Additionally, ACCA (The Association of Chartered Certified Accountants) and Titanium Media will jointly release a thematic report during the session, sharing the latest trend analyses.

Special Closed-Door Session 3 | From Component Precision to Production Line Intelligence: Capturing AI-Driven Industrial Upside

AI compute infrastructure is accelerating rapidly, continuously reshaping every link in the supply chain. In the face of massive demand and transformation, how can manufacturing create greater value? What significant changes are occurring within supply chains? How are various stages adapting?

This session, co-organized with Industry Innovation Alliance & ITES Shenzhen Industrial Expo, invites leading companies across the AI compute supply chain—from data centers to liquid cooling, materials, and manufacturing methods—to explore the novel transformations and challenges underway.

On-site, representatives from both supply and demand sides—including senior executives from prominent OEMs—will join forces with supply chain leaders like Siemens, Taigang Stainless Steel, and Cold Spring Energy Control to discuss and clarify: What kind of solutions are most urgently needed by demand-side stakeholders during rapid AI iteration? How can supply chains evolve from “capable of production” to “stable mass delivery”? And how can each segment of the industrial chain capture the value of the AI era?

Special Closed-Door Session 4 | Physical AI Enters Reality: Bridging Real Needs with Cutting-Edge Technology

AI deployment in the physical world remains one of the most discussed and debated topics in the industry today. From flashy demos to actual production-grade applications, how far are we really? How do we identify viable scenarios and integration points?

This closed-door session, co-hosted with Beijing Academy of Artificial Intelligence, brings together industry practitioners and technology providers to find synergies between real-world needs and physical AI capabilities, exploring mutually beneficial pathways forward.

Yang Hui, General Manager of Yuexiu Property’s Digital Intelligence Development Center, and Li Shaochun, Digital Strategy Scientist & CIO of Yunnan Baiyao Group’s Yunbai Tech, will present industry perspectives on physical AI implementation in real enterprise contexts. Wang Pengwei, Head of the Embodied AI Model Research Center at Beijing Academy of Artificial Intelligence, and Liu Dong, Founder & CEO of StarYuan Intelligent Robotics, will present technological advancements in world models and embodied intelligence, discussing the current boundaries of physical AI and how it can drive industrial upgrading.

Special Closed-Door Session 5 | How to Become an AI-Native Super Organization

Today, enterprise AI pilots and Agent employees are proliferating, yet scaling meaningful value remains elusive. The root cause may lie in the classic dilemma: “easy to introduce technology, hard to reform organizational structure and management mechanisms.” In this context, “AI-native” is increasingly seen as the solution—but truly integrating AI into the intricate details of operations and management is no simple task.

This session dives into real organizational management environments, exploring feasible paths from “applying AI” to becoming an “AI-native super organization.” Enterprise and industrial leaders will share and discuss their philosophies and practices in building AI-native organizations. For instance, Zhang Bin, Director of Solutions at DIPU Technology, will present practical implementations of the enterprise Agent ontology model, demonstrating how Agents evolve from “thinking” to “understanding factories”; Cong ZeLong, AI Product Center Director (Gen Z) at Pinsheng, will share his company’s internal journey toward AI-native transformation.

Special Closed-Door Session 6 | How Al Coding Can Deliver True Productivity

Coding is the first frontier for AI commercialization. Over the past year and a half, Anthropic has led a dramatic leap in model-based AI coding capabilities—this wave continues to surge without abating. Meanwhile, an increasing number of coding products are emerging and iterating, opening up new market spaces.

This closed-door session focuses on how Al Coding can be embedded into enterprise workflows to generate true productivity. Chu Qing, Founder and Chairman of Zhishi Shen Gong, will trace the historical evolution and emerging trends of Al Coding; Li Bin, Technical Lead of Alibaba Cloud Qoder, will share practical insights from a core player’s perspective on what it means when Agents become colleagues and how AINative organizations operate; Zhang Chaohui, digital strategy expert and former Chief Information Officer of Ping’an Securities, will draw from over thirty years of coding and code management experience to examine the real challenges organizations face in the “AI programming era”: How should enterprises leverage AI Coding effectively? In an age where AI-generated content and code are rampant, what constitutes real value—and where does it come from?

Billion-Dollar Club Thinkers’ Session

AI is not merely another round of digital tool upgrade—it represents a generational leap in enterprise organizational forms and technological systems. State-owned enterprises, central SOEs, and leading private firms must collectively define “why we must act” at the strategic level.

Currently, large-scale enterprise AI deployment faces widespread challenges: high investment, scattered returns, and difficulty in measurement. At the institutional level, how do we establish organizational mechanisms, management processes, and governance frameworks suited for AI? At the value level, how do we quantify AI value, validate business effectiveness, and close the input-output loop? Titanium Media, in collaboration with enterprise digitalization pioneer Chaojuhe, has invited select guests to co-host this thinkers’ session at the Digital Value Conference, aiming to systematically explore and respond to the two core concerns: “How to prove value?” and “How to break organizational resistance?”

The grand exam of AI system deployment tests not just AI itself, but everyone involved. From Token economics to corporate accounting, from business scenarios to organizational structures, from compute to coding, this full-day series of closed-door sessions will form a practical, actionable roadmap and guidebook for AI implementation, preparing for the myriad challenges and uncertainties ahead.

AI is still evolving. On September 16–17 in Sanya, join top decision-makers in keeping pace with the times.

Source: Flying to TAI Space