The tech industry is ablaze with activity once again in the opening month of 2026.

These days, the typically reserved Pony Ma has ignited fervent discussions by openly outlining Tencent's AI strategy—a rare move for him. Even more uncommon is the public emergence of Jack Ma, a business magnate who has maintained a low profile for an extended period, to discuss AI and education. Even Cheng Wei, known for his reticence, has been making frequent public appearances, engaging in AI dialogues with the president of Tsinghua University.

In the public eye, Pony Ma is seen as an introverted figure who seldom voices his opinions. Jack Ma, though still a beloved figure, has kept a low profile in recent years, rarely making public appearances and speaking out only occasionally, and even then, mostly within the confines of Alibaba.

The public appearances of both Mas to discuss AI carry profound implications: In today's era of technological saturation, AI innovation is entering a critical "sprint" phase.

Why are only the tech behemoths positioned at the starting line of this AI sprint?

Why are Pony Ma and Jack Ma making frequent appearances at this juncture?

I believe a pivotal reason may be that, at this crucial juncture of industry transformation, there is heightened anticipation for Alibaba and Tencent to spearhead more distinctive AI innovations.

With a major showdown looming, the key lies in fostering consensus. In the AI sprint phase, the true test may not be the complexity of technological innovation or the magnitude of resource investment, but rather the mindset towards AI innovation.

On one hand, in recent years, the entire tech industry has become more risk-averse in its approach to innovation, displaying a diminished appetite for bold experimentation.

Nowadays, when large corporations launch new ventures, they no longer boldly experiment but instead meticulously consider ROI, budgets, and even consider merely avoiding cuts to their innovation departments as a success.

When it comes to AI, big companies may also harbor fears of risk.

Just as Wu Yongming announced Alibaba's 300 billion yuan investment in AI, some labeled it a "colossal gamble." When Pony Ma declared Tencent's pursuit of AI social networking, it was interpreted as "Tencent's anxiety." Frankly, the market is inherently "profit-driven and risk-averse," instinctively pursuing "duty" and shunning risk.

At this juncture, there is a pressing need for influential voices to step forward and provide reassurance for innovation.

From the perspectives of Jack Ma and Pony Ma, AI necessitates breaking free from existing "comfort zones" and embracing some "unconventional" maneuvers, venturing into areas that may seem "unrelated to the core business."

Elon Musk ventured into "unrelated" territories, founding PayPal for profit before pivoting to rockets, leading to the creation of SpaceX. Liang Wenfeng, despite being "unrelated" to his core business, led a quantitative trading company to develop DeepSeek, thereby altering the trajectory of China's AI development.

Today, when tech giants pursue AI innovation, they predominantly build upon their existing capabilities, failing to truly break free from established frameworks.

For Alibaba, ByteDance, and Tencent, AI is not a "new business" but an extension of their existing competencies.

Take Qianwen AI as an example. Initially, Qianwen's role within Alibaba was not to vie for a super entry point but to serve as a new technology experiment. Its reputation was built not among C-end users but within Alibaba's open-source community.

It wasn't until November of the previous year that Qianwen shifted towards a B2C model. Subsequently, Alibaba established the Qianwen ToC Business Group, and various businesses within the Alibaba ecosystem, including Shangu and Fliggy, began integrating Qianwen.

Thus, Qianwen AI may also be a product of Alibaba's "B2B to B2C" logic, reflecting Alibaba's typical entrepreneurial approach.

From an operational standpoint, at Alibaba's Qianwen AI launch event, Wu Jia used Qianwen to order milk tea. As an AI entry point, Qianwen's potential to restructure Alibaba's fundamentals became evident. Alibaba's core lies not in content but in its supply chain and service capabilities, so Qianwen AI can serve as both a traffic entry point and a "service entry point."

Under this logic, maximizing the value of Taobao and Ant Group's service ecosystems under Qianwen AI's leadership may be a pivotal path for Alibaba's AI implementation.

The situation is similar for ByteDance.

Doubao AI's accomplishments are remarkable, especially after securing a partnership with the Spring Festival Gala.

However, at its core, Doubao AI is no different.

In a sense, Doubao AI emerged from ByteDance's "APP factory" capabilities, extending its ability to create hit products.

The success of Douyin and TikTok has drawn attention to ByteDance's algorithmic prowess, overshadowing its "APP factory" capabilities. Before the AI boom, two internet apps with over 100 million daily active users—Hongguo and Fanqie—were both ByteDance products.

At the beginning of 2025, Doubao gained popularity. Initially, videos of users "voice-training" Doubao went viral on Douyin, followed by the development of various Doubao AI playstyles. Coinciding with the explosion of AI multimodal capabilities, Doubao AI soared.

Doubao AI's success lies not only in ByteDance's algorithmic strength but also in the "AI factory paradigm" upgrade. The APP factory has transformed into an "AI playstyle factory." Insiders at Doubao reveal that many of its playstyles were not driven by top-down decisions but by proactively seeking out scenarios and playstyles.

Thus, today's Doubao AI is another super traffic entry point, an "AI version of Douyin."

In the AI sprint phase, Doubao and Qianwen have reversed roles. Doubao lacks no traffic but still needs supply. The fact that Doubao AI phones were banned by various major companies indicates that commodity and service supply remains a point to address.

AI innovation in the era of technological saturation is not just about traffic and promotional capabilities but also about scale and infrastructure supply capabilities.

For ByteDance, achieving the top market share with Doubao may not be arduous, but that is just the baseline. How to profoundly enhance commodity and service supply and integrate it into Doubao AI's ecosystem may require billions or even tens of billions in investment. This is likely a key focus for Douyin going forward.

ByteDance possesses the confidence to build supply chains. On one hand, ByteDance has made few external investments, leaving it well-funded. On the other hand, ByteDance is still young and dynamic. How to further unleash this dynamism in the new AI sprint phase depends on the surprises Doubao can deliver.

Beyond Doubao, the recent focus has been on Yuanbao AI.

Tencent taking the lead in creating a new AI social product is not surprising but rather an expected yet reasonable move, similar to how, during QQ's heyday, Pony Ma still had Zhang Xiaolong develop WeChat.

AI social products also align with Tencent's existing capabilities.

Pursuing AI social networking may seem like Tencent is "sacrificing its core" for "self-disruption," but it is actually "buying insurance." Tencent's foundation lies in "social + advertising + gaming," with social being the cornerstone—the "1" before countless zeros. Thus, AI's greatest value for Tencent lies in preventing disruption.

I boldly predict that the success of Tencent's AI social product may not be the most critical factor. What matters is that the social cornerstone remains undisrupted. On this basis, an additional attention entry point could mean another monetizable traffic pool.

Tencent with Yuanbao AI, Ant Group with Qianwen, and ByteDance with Doubao are all venturing into the "uncharted territory" without hesitation.

Whether they can emerge from this uncharted territory and forge their own AI path remains to be seen. However, with tech giants doubling down, the AI industry's competition is set to intensify. Second- and third-tier large model makers may face even tougher times ahead.

Tech Giants Step Up, Mid-Tier Models Face the "Elimination Line"

With ByteDance, Tencent, and Alibaba all entering the fray, the AI industry is about to enter its next transformation phase, with mid-tier large model makers likely suffering the most.

While big companies have reached the "sprint" phase, the "elimination round" for mid-tier AI players is far from over.

Data from the Tianyancha APP shows that, as of January 29, 2026, among active or operating AI-related enterprises, approximately 869,000 were established within the past 1–3 years. Around 168,000 AI-related companies were registered as dissolved, revoked, or in liquidation.

Recently, a popular blogger named "Lao A" introduced the concept of the "U.S. Elimination Line." Some have hailed it as a significant achievement in the public opinion arena in decades.

In this era of profound transformation, more and more AI companies may find themselves on the "AI elimination line."

Today, AI companies can be roughly categorized into three types:

The first category includes AI businesses under tech giants like ByteDance, Tencent, and Alibaba, such as Yuanbao and Qianwen.

These AI players are farthest from the elimination line.

The second category consists of "mid-tier players."

These are AI products from second-tier internet and tech companies, such as 360 Nano and iFLYTEK Spark.

These large model players often do not solely focus on general-purpose models but instead seek alternative paths. Whether they can survive depends on whether they can further integrate with their existing businesses. If they can fuse with their original businesses, new opportunities may arise.

Otherwise, they may inch closer to the "elimination line."

The third category comprises independent large model companies.

Independent large model companies fall into two subtypes: general-purpose AI large models, such as Zhipu AI and Yuezhi Anmian.

One prediction is that as big companies concentrate their investments in general-purpose AI large models, market share may become increasingly solidified, leaving less room for independent large models.

Moreover, from a commercial perspective, AI large models are crucial entry points and must-win areas for big companies. Unless independent large model players can be acquired by big companies, it may not be a lucrative business. After all, independent large model companies lack both funding and ecosystems. When the industry focuses on scale, cost, and commercialization, these players' survival environments may become increasingly challenging.

Here's a "bold claim": If independent large model companies cannot achieve DeepSeek-like innovation within the next year, some may face even greater difficulties and even trigger the "elimination line."

The other subtype is vertical large models.

Examples include Kunlun Wanwei's "Tiangong" large model and Wondershare's "Tianmu" large model. These companies may be even closer to the "elimination line."

When it comes to commercialization, these players may face a "double bind."

First, ToB vertical domains may still face direct competition with tech giants.

For instance, developing music large models may inevitably lead to competition with Tencent, while video large models may face platforms like Kling and Jimeng, and ToB large models may encounter Baidu.

Second, even if they secure a market position, they may face questions about scale effects.

An interesting trend is that OpenAI and Google's recent product launches highlight a key industry trend: While giants innovate, they also streamline structures and optimize costs.

This means that the AI arms race among the industry's leading players is shifting from pursuing absolute performance to pursuing cost-efficiency.

As the industry gradually pivots, will vertical large model players maintain cost and efficiency advantages? This is a critical question.

If Lao A's "elimination line" is chronic and structural, then the AI-driven industrial "elimination line" may be rapid and disruptive. After all, the market will not wait for slow bleeding but will directly disrupt rules and reshape the landscape.

However, in the AI era's innovation race, there are no eternal giants—only eternal innovation.

DeepSeek's overnight disruption of ChatGPT, Tencent Yuanbao's "overnight comeback" in market share, and ByteDance Doubao AI's "latecomer advantage"—these stories of AI innovation will continue to unfold.

The script remains the same, but who will write this AI innovation era?

We shall see as we move forward.

Disclaimer: This article is based on legally disclosed company information and publicly available data, with commentary provided by the author. However, the author does not guarantee the completeness or timeliness of this information. Additionally, the stock market carries risks, and caution is advised when entering. This article does not constitute investment advice, and investment decisions must be made independently.