In mid-2026, TCL’s Chairman Li Dongsheng attended the Davos Forum, where he repeatedly highlighted AI exploration within the upstream supply chain.

Before his remarks faded from memory, TCL, at the corporate level, outlined a clear strategic vision for its two major capital-intensive businesses under TCL Technology. Facing the AI era, the company aims to build capabilities around 'interaction gateways' and 'energy foundations,' transforming these into long-term growth drivers.

Screens act as gateways for interaction, while photovoltaics supply energy for computing power. Li has positioned these two long-standing businesses as key engines for the next growth cycle.

Looking back at TCL’s expansion over the past two decades, Li Dongsheng has consistently leveraged heavy asset investments to gain a competitive edge. To expand globally, he spearheaded the acquisition of Thomson’s TV business; to master core display technologies, he championed the construction of CSOT; and to broaden terminal operations, TCL continuously expanded its product range through both internal development and acquisitions.

These strategic moves have brought new technologies, markets, and operational capabilities to TCL. However, they also subjected the company to lengthy construction cycles and the inevitable costs associated with misjudgments and integration challenges.

As we look ahead, can an entrepreneur accustomed to securing industrial positions through heavy asset investments continue to replicate past successes in an era where value increasingly shifts towards software, AI, and user relationships?

As a leader in traditional manufacturing, Li Dongsheng and TCL find themselves in a unique position in the AI era.

TCL’s core portfolio comprises three main segments: TCL Technology, focused on semiconductor displays, reported revenue of RMB 88.65 billion and a net profit attributable to shareholders of RMB 3.81 billion in the first half of 2026, nearly doubling year-on-year. TCL Electronics, which handles smart terminals and internet services, reported revenue of HKD 63.76 billion and a 40.2% year-on-year increase in net profit attributable to shareholders to HKD 1.53 billion during the same period. TCL Zhonghuan, rooted in photovoltaics, reported a net loss of RMB 3.2 billion but has reduced losses for two consecutive quarters.

Regarding the link between upstream operations and AI, Li Dongsheng, drawing on his industrial background, has crafted a strategy centered on interaction and energy. Under this framework, TCL Technology and TCL Zhonghuan are experiencing vastly different operational realities.

Downstream, Li Dongsheng appears to have a wealth of options—TVs, air conditioners, refrigerators, washing machines, mobile phones, glasses. TCL’s terminal portfolio covers nearly every aspect of home life. Many of these capabilities were not built specifically for the current AI boom but provide the foundation for developing new products and experimenting with cross-device services today.

In the past, Li Dongsheng expanded TCL’s capabilities with each investment. Now, AI presents a new challenge to his management approach: Can these separately accumulated capabilities work together to create businesses that were previously unattainable?

I. How to Sustain 'Crude' Advancement?

Time is like a magician, once allowing crude advancements to outperform refined backwardness. But for entrepreneurs who make early bets, time also means continuous investment and the cost of waiting for returns.

To some extent, Li Dongsheng adheres to this industry-oriented positioning logic. Within a large cycle, as long as a company makes relatively correct judgments, it can identify segments with sufficient scale, high entry barriers, and key positions in the industrial chain, thereby exchanging capital, manufacturing, and efficiency for a strategic foothold.

This approach has not become obsolete with the advent of AI.

Generative AI’s expansion from mobile phones and computers to TVs, automobiles, glasses, and other terminals has significantly increased the scenarios for human-machine interaction. Screens no longer just display content; they also handle information feedback, visual interaction, and multi-device collaboration. TVs are getting larger, cars are incorporating more screens, and glasses are bringing displays even closer to the human eye.

CSOT, which nearly supports TCL Technology’s entire business, is well-positioned on this industrial chain.

In the first half of 2026, TCL CSOT reported revenue of RMB 50.27 billion and a net profit of RMB 3.9 billion, with net profit attributable to TCL Technology reaching RMB 3.28 billion, up 24.8% year-on-year.

'Small markets cannot meet the growth needs of large enterprises,' Clayton Christensen wrote in The Innovator's Dilemma. It should be noted that this round of profitability is more a result of supply-side restructuring.

TCL early on acquired LCD production capacity from Guangzhou's 8.5-generation line. With TV sizes increasing in recent years, LCD has gained more downstream demand and orders due to its higher cost-effectiveness compared to OLED. Steady demand expansion has absorbed capacity, while the accelerated exit of Japanese and Korean manufacturers has also helped shift the supply-demand balance.

At a fundamental level, TCL can be said to have waited for the right moment.

Meanwhile, TCL continues to advance the construction of high-generation printed OLED production lines and extends its products from TVs to more scenarios such as laptops, automotive displays, and professional displays. For example, TCL CSOT’s single-chip full-color silicon-based MicroLED solution released last year targets the AI glasses sector, where investment is already flowing.

The more dispersed AI terminals and diverse display formats may find more outlets for the production lines and technical reserves TCL has built through heavy investments in the past.

Using LCD’s cyclical dividends to provide a window for panel supply to penetrate into AI-driven new hardware forms the A-side of Li Dongsheng’s AI narrative.

Screens correspond to interaction gateways, and photovoltaics correspond to energy foundations. Under the same narrative, panels are enjoying the tailwinds of supply consolidation, while photovoltaics are facing the headwinds of oversupply. The actual performance of heavy asset investments inevitably encounters cyclical resistance.

AI training and inference require significant electricity, and data center construction has indeed raised global energy demand. Understanding photovoltaics as the energy foundation behind AI is logically sound in the long term.

But demand growth does not directly translate into profits, bypassing industrial cycles. In the first half of 2026, TCL Technology's New Energy Photovoltaic and Other Silicon Material Business Segment reported revenue of RMB 14.31 billion and a net loss of RMB 3.457 billion.

In terms of financial structure, the cost pressure from photovoltaics is more straightforward. Depreciation begins annually from the day a silicon wafer production line is built, unaffected by price fluctuations. For example, TCL Zhonghuan recognized asset impairment provisions of RMB 4.622 billion in 2025, including RMB 3.185 billion in inventory write-downs, and an additional RMB 526 million in the second quarter of this year.

Under the holding costs of heavy assets, Zhonghuan continues to experience industry adjustments driven by supply-demand imbalances and price pressures. Long-term energy demand growth can coexist with the immediate supply-demand imbalances and price pressures faced by manufacturers.

Just as the cycles of panels and photovoltaics are misaligned amid the AI wave.

II. The Permutations and Combinations of Full Category Operations

Having endured the hardships of cyclicality, Li Dongsheng understands better than anyone the category constraints of traditional manufacturing. Thus, using a combination of multiple categories to hedge against fluctuations in a single segment becomes a logical strategy.

The value of full category operations can be understood at two levels. On the surface, diversified revenue can disperse some operational risks. More importantly, this strategy allows different businesses to leverage each other, reusing brand and channel momentum while further forming hardware gateways for the AI era.

Historically, Li Dongsheng has been more adept at driving vertical integration along the industrial chain, exemplified by CSOT.

Extending from TVs upstream to connect core components, manufacturing, and terminals, cost, performance, and delivery form clear bases for collaboration. Today, TVs, air conditioners, and glasses each have their own industrial chains. To serve the same user across these products requires further horizontal coordination among multiple industrial chains.

Judging by assets alone, TCL’s fundamental position is impressive.

The interim report shows that in the first half of the year, TCL TV’s international market revenue reached HKD 25.44 billion, up 29.6% year-on-year, accounting for over 70% of TV business revenue. The foundation of international revenue stems from the costly acquisition of Thomson in 2004, which laid the groundwork for the group's cross-border operations, channels, and organizational capabilities.

In terms of categories, years of acquisitions have also established a solid foundation. On July 15 this year, TCL announced a planned acquisition of a 51% stake in TCL Air Conditioner Business for HKD 5.61 billion, expected to be completed in the fourth quarter of 2026, thereby reincorporating white goods into the listed entity.

The more substantial the assets, the more critical the judgment in betting decisions. From TCL’s exhibition at the recent IFA, the company appears to aim for 'saturation coverage' of users’ diverse needs inside and outside the home.

On display were mobile phones, tablets, glasses, air conditioners, refrigerators, washing machines, security products, and more, all unified under the theme of 'Inspired Living.' Traditional home appliances and modern portable terminals are both within the group's reach.

Such a product lineup showcases TCL’s industrial accumulation but also brings the challenges of full category operations to the fore.

Peter Lynch once used the term 'diversification disaster' to mock companies that blindly expand, especially by spending profits on overpriced acquisitions beyond their understanding. Thanks to its past accumulation, TCL does not appear to face such troubles on the surface and can even explicitly promise a 30% cash dividend to investors. However, in the AI era, the challenge of diversification lies in the increasing number of businesses vying for capital and management attention simultaneously.

Since AlphaGo ushered in the intelligent era, the evolution of hardware intelligence has followed three clear axes: TVs, air conditioners, and washing machines serve as platforms for home entertainment and life services; portable terminals like mobile phones and glasses strive for more frequent interaction; and health monitoring functions in body fat scales, fitness bands, and rings attempt to provide continuous services based on bodily data.

At the software and system levels, there may be overlaps. However, each demand has a distinct product logic behind it. Take air conditioners, for example: the core competitiveness of high-end air conditioners has long revolved around silence, energy efficiency, temperature control, and channel services. AI struggles to provide clear improvements in consumer experience, making commercial returns elusive. As for screens, which the group has heavily invested in, they appear to be among the few winners in this AI hardware competition.

Broaden the perspective, and AI has not unified the hardware world as narratives suggest; instead, it has made hardware forms even more fragmented. Li Dongsheng has used capital and manufacturing capabilities to position TCL at key points along the industrial chain. Now, AI’s interaction demands compel TCL to explore more at the software level.

The interim report shows that TCL’s internet business, reflecting its software capabilities, reported revenue of HKD 1.69 billion in the first half of the year, accounting for about 2.6% of TCL Electronics’ revenue. This indicates that building a unified intelligent interaction layer and a 'living habitat' is no easy task.

Vertical integration relies on control, while horizontal integration relies on coordination. The former is a traditional strength of manufacturing, while the latter is more akin to the work of software companies.

III. The Limits of Li Dongsheng's Experience

Coordination is not a new challenge for Li Dongsheng. Over the past two decades, his choices have included both acquiring businesses and divesting them. Securing an industrial position requires investment, and he has also paid tuition in managing those investments afterward.

In 2006, amid the post-Thomson acquisition predicament (dilemma), Li Dongsheng wrote The Rebirth of the Eagle. When asked whether he would still pursue those two cross-border acquisitions, his answer was, 'Definitely.' However, in his reflection, he also criticized internal favoritism and trust issues.

The assets TCL has today stem not only from his willingness to invest early but also from the process of repairing operational capabilities after investment setbacks. A notable example is the 2019 TCL restructuring, where TCL Industries took over terminal operations.

By 2026, this division of labor has also occurred within Li Dongsheng himself. On January 19, upon Li Dongsheng's nomination, the TCL Technology board appointed Wang Cheng as CEO, responsible for overall daily operations. Li Dongsheng continued as chairman, relinquishing the CEO role.

A pure industrial perspective has its limits, especially in today’s rapidly changing consumer market. Upstream thinking can calculate costs, performance, and capacity but cannot grasp user preferences and retention.

Wang Cheng, who joined TCL in 1997, has experience in overseas business management and group human resources management, later serving as CEO of TCL Electronics and TCL Industries. His career spans the boundary between consumer-facing terminals and enterprise-facing upstream industries.

Li Dongsheng may find it difficult to break free from the inertia (inertia) of traditional manufacturing thinking, making Wang Cheng a 'satisfactory solution' for the group at present.

Following this line of reasoning, it becomes readily apparent that since Wang Cheng assumed the role of CEO, TCL has progressively adopted a software-centric orientation. The interim report reveals that TCL Technology's overall investment volume in the first half of 2026 witnessed a 19.34% year-on-year decline. As industrial investments contract, the share of 'soft power' starts to ascend.

At this year's IFA event, TCL unveiled the Turing platform, which amalgamates models, generative AI services, and multimodal intelligent agents into a unified technical framework. This platform supports a range of products, including air conditioners, refrigerators, washing machines, and more, through the Smart Brain. This move at least suggests that TCL is pursuing capabilities that can be shared across multiple business units, rather than just expanding its range of terminal devices.

Every stride Li Dongsheng took to 'advance' in the past contributed an industrial asset to TCL's portfolio. Now, by taking a strategic step back, he has created more room for the group's day-to-day operations and enabled himself to focus on areas where his expertise remains valuable: observing industry trends, economic cycles, and strategic directions.

For an entrepreneur who measures success by industrial standing, acknowledging the limitations of experience may represent the ultimate and most challenging industrial decision he must confront.