Source | YuanSight

In the wee hours of March 6, the tech world experienced a seismic shift.

An omnipotent AI agent named Manus stormed social media, trending across multiple platforms. On secondary marketplaces, an invitation code fetched an astonishing RMB 50,000; in the A-share market, AI agent-related stocks witnessed a surge of daily limits.

Screenshot from Manus's official website

Amidst this frenzy, the fate of a venerable tech giant hangs in the balance - Lenovo, the world's largest PC vendor, despite embracing the AI era with AI PCs, now faces formidable challengers like Manus.

Manus not only showcases technological prowess but also disrupts the traditional hardware upgrade paradigm of PC manufacturers. As AI takes over computer operations via browsers, the race for local NPU chip computing power suddenly appears futile.

Perhaps, the physical constraints of hardware are being transcended by the invisible hand of algorithms.

01

The Cloud Revolution

The emergence of Manus rips off the final veil of AI PCs.

Developed by a Chinese team, this universal agent can autonomously tackle complex tasks such as financial report analysis and poster design simply by leveraging the Chrome browser and OpenAI interface. Its operations are entirely reliant on cloud computing power, with minimal requirements for local hardware.

According to feedback from beta users, Manus's cloud-based solution surpasses local deployment by 23% in image generation speed, while costing only 17% of the latter. This experience shatters the "computing power obsession" narrative propagated by traditional hardware manufacturers - when users discover that Manus runs seamlessly on a 2018 MacBook Pro, the rationale for hardware iteration is called into question.

In contrast, strategies of traditional PC manufacturers like Lenovo continue to revolve around hardware upgrades. For instance, Lenovo's launch of the world's first rollable screen AI PC, the ThinkBook Plus Gen6, boasts impressive flexible screen technology: achieving screen extension at a thickness of 19.9mm, supporting offline AI functions, and equipped with an Intel Core Ultra processor. However, the production costs cannot be overlooked - the starting price is as high as $2999, far exceeding mainstream consumer-grade PCs, hindering the widespread adoption of AI technology among the general public.

Image source: IT Home

This innovation, indifferent to cost, is interpreted within the industry as an extreme manifestation of "hardware anxiety." Besides Lenovo, ASUS has launched the "lightest Copilot+PC," HP has upgraded the OMEN MAX gaming laptop, and Samsung has showcased the AirJet cooling technology concept machine. Yet, all these attempts confront the same dilemma: morphological innovation fails to compensate for the lack of cloud services.

Indeed, the mismatch between data and user perception of AI PCs is becoming apparent. According to market research firm Canalys, in the second quarter of 2024, shipments of Windows PCs equipped with AI technology surged by 127% quarter-on-quarter. During this period, Lenovo increased its share of AI Windows PC shipments to about 6% with products like the Yoga Slim 7x and ThinkPad T14s, which feature the Snapdragon X (AI PC chip).

However, user feedback on social platforms such as Xiaohongshu indicates that most consumers have a vague understanding of AI PCs' core functions, such as local large models and heterogeneous computing power fusion. One technology blogger bluntly stated: "So-called AI PCs are more like high-end computers with a few pre-installed AI tools."

02

The Decline of Local Computing Power

Lenovo's dilemma is intrinsically tied to its technological route judgment. Its heavily promoted AI PC concept is built on the premise that "local computing power determines the AI experience," overlooking the evolution of agent workflows.

Judging from the current evolution speed of large AI models and agents, the day may come when Qualcomm unveils a high-performance processor, and the developer community will focus on "whether it can natively run agents like Manus." This shift in value evaluation criteria makes the PC upgrade narrative less compelling.

While AI PCs have driven an increase in Lenovo's PC average selling price, the operating profit margin of its smart device business has hovered around 7.3%. In contrast, NVIDIA's gross margin has remained around 70% for the same period, according to its financial report. This contrast affirms industry observations: the value of AI is shifting from terminal hardware to the cloud.

Lenovo has attempted to break the deadlock with a "hybrid cloud" strategy, introducing devices like the YogaPad Pro AI Yuanqi Edition to establish an "individual intelligent twin" ecosystem. However, its software business revenue accounts for less than 8%, lagging significantly behind ecosystems like Huawei's HarmonyOS and Xiaomi's Mi Share.

Furthermore, Lenovo has unveiled multiple new AIPC products simultaneously. However, the embarrassment lies in their pricing, with most exceeding RMB 10,000, limiting user coverage to the high-end market. In contrast, Huawei and Xiaomi have constructed a more comprehensive AI ecosystem through the full-device collaboration of mobile phones, PCs, and tablets.

Microsoft will terminate support for Windows 10 in October 2025, and this anticipated influx of hundreds of millions of device replacements is becoming the final "race against time" for AI PCs.

In fact, Lenovo's ecosystem strategy exhibits internal and external fragmentation - the integration of its internally developed "Tianxi AS" agent system with large models like Alibaba's Qianwen and Baidu's ERNIE remains at the API call level, failing to form deep collaboration. This "closed ecosystem + external grafting" model not only stifles developer creativity but also increases the difficulty of responding to challenges from open-source ecosystems like Manus.

03

The Ecosystem Game

The Manus phenomenon uncovers a harsh truth: in the AI era, hardware manufacturers are transforming into "OEM factories" for algorithm giants.

As the AI agent movement to "squeeze the performance out of PCs" progresses, the traditional 3-5-year upgrade cycle for PC hardware may be indefinitely extended. In other words, AI agents, through overclocking and other techniques, are revitalizing many older PCs, directly impacting Lenovo's basic replacement rate.

The deeper contradiction between the two lies in their business models. Hardware sales rely on one-time revenue, whereas the subscription model for agent services is more sustainable. As Manus introduces services with monthly fees ranging from $50 to $100, Lenovo is still grappling with a hardware profit margin of around 7.3%. This mode conflict is shaking the foundations of AI PCs.

As developers create 12 vertical scenario plugins for Manus, such as resume optimization and contract review, the "light photo editing" and "voice transcription" functions of AI PCs may become transitional products in technological evolution.

Thus, the outcome of this competition may already be predetermined. When Huawei's HarmonyOS PC achieves 20 TOPS of local computing power + 200 TOPS of elastic cloud expansion through end-cloud collaboration, and NVIDIA's RTX 5090 graphics card offers 4000 TOPS of computing power, Lenovo's adherence to the X86 architecture + NPU solution is becoming the "boundary" for technological innovators.

History does not repeat itself, but it rhymes. Just as Nokia fell in the smartphone era, the revolution sparked by AI agents may be penning a new chapter for the PC industry.

Lenovo's transformation dilemma fundamentally stems from the structural contradiction between its heavy asset model and lightweight innovation. Its vast system of over 30 manufacturing bases worldwide appears bloated in an era dominated by algorithms. In contrast, the Manus team, with just over 200 people, can complete the entire process from prototype development to ecosystem construction within six months. This agility gap reflects the deep institutional issues of traditional hardware manufacturers.

The explosion of Manus is not an endpoint but a starting point. This Chinese AI product not only showcases technological breakthroughs but also signifies the redistribution of industrial power. As algorithms begin to define hardware value, the transformation window for companies like Lenovo may be shorter than anticipated.

In this multi-dimensional war of computing power, ecosystems, and business models, the only certainty is uncertainty. The only thing that remains certain is that players who dare to reconstruct the value chain will ultimately reap the rewards of the era.

Some images are sourced from the internet. Please inform us if there is any infringement for removal.