Cheetah Mobile (NYSE: CMCM) is set to announce its Q2 2026 financial results on September 11. Historically, it may still be perceived as an established internet company in transition: its traditional advertising business faces pressure, investments in robotics and AI have yet to yield full returns, and overall profitability has not yet reached ideal levels.

However, if the focus shifts from short-term profit and loss to revenue composition, a different picture of Cheetah Mobile emerges.

In Q1 2026, Cheetah Mobile's revenue from cloud and AI infrastructure services reached RMB 46.8 million, up 68.3% year-on-year, accounting for 18.1% of total revenue. Revenue from robotics and other businesses soared to RMB 51.2 million, up 175.9% year-on-year, representing 19.8% of total revenue. Together, these two emerging businesses contributed nearly 38% of the company's revenue. Management expects their combined revenue share to exceed 50% in the second half of 2026.

This suggests that the key question for the upcoming Q2 results may not be whether Cheetah Mobile can "strike gold" in AI, but whether the market continues to evaluate the company based on the framework of a traditional internet company, despite its clearly transformed revenue sources.

I. New Growth Drivers Masked by Legacy Businesses

Cheetah Mobile's total revenue in Q1 was RMB 259 million, roughly flat year-on-year. At first glance, the company appears stagnant. However, a closer look reveals that this stagnation is primarily due to drag from legacy businesses, while new growth drivers have emerged.

In Q1, revenue from internet services declined 15.2% year-on-year, and global enterprise services revenue fell 10.5%. The latter decline was mainly due to policy changes affecting overseas advertising platforms, leading to a 51.5% drop in ad agency service revenue. Meanwhile, cloud and AI infrastructure revenue within global enterprise services grew 68.3% to RMB 46.8 million, accounting for 64.3% of segment revenue.

This highlights a shift that is easily obscured by total revenue figures: the decline in legacy businesses offsets growth in new ones, but the quality and value sources of revenue within the company have fundamentally changed.

If cloud and AI infrastructure revenue is simply annualized based on Q1, it would reach approximately RMB 187 million. While this is not a full-year forecast, it demonstrates that this is no longer a conceptual business stuck at the strategy announcement or product demo stage. More importantly, Cheetah Mobile disclosed that platform-wide daily Token usage surged more than 20-fold from January to May 2026, exceeding 400 billion in May.

Token volume alone does not equate to revenue, nor does it directly translate to profit. Declines in model prices, shifts in client mix, and varying service combinations can all cause Token usage and revenue to move out of sync. However, rapid Token usage growth at least indicates that Cheetah Mobile has entered the actual deployment phase of enterprise AI resources. Meanwhile, cloud and AI resource usage is inherently sustainable, requiring long-term services such as billing management and technical support during subsequent usage.

II. As AI Models Proliferate, Enterprises Need an 'Intermediate Layer'

Token growth ultimately depends on whether AI can integrate into real production workflows. Coding has already validated the commercial value of high-frequency AI calls, while sales, customer service, operations, and other enterprise scenarios are still in early penetration stages. If more AI Agents can deliver clear ROI, enterprise Token consumption and related cloud resource demand could see longer-term growth.

At the same time, how enterprises use AI is evolving. Over the past two years, industry attention has focused on model capabilities and compute supply. However, as the number of models rapidly increases and prices continue to shift, new complexities arise when enterprises actually deploy AI in production environments.

Multi-cloud is not a new concept. Enterprises already adopted multi-cloud architectures during the cloud era, and the AI era has added a new dimension of complexity: multi-model management. Enterprises must now manage not just "which cloud hosts which workloads," but also "which tasks use which models, run on which AI infrastructure, and how to balance effectiveness, cost, and stability." Model access, cost, stability, and usage management have become new enterprise needs.

Cheetah Mobile's Cloud & AI Infrastructure business sits at this supply-demand nexus: one end connects Chinese companies going global, the other connects overseas cloud and AI resources from AWS, Google Cloud, Microsoft, and others, while gradually adding multi-model access and management capabilities.

It is important to avoid an overly simplistic understanding: Cheetah Mobile's value is not just procuring cloud or Token resources upstream and providing them to clients. Enterprises could, of course, purchase directly from cloud providers. However, actual cloud and AI resource usage, especially for large enterprises, generates substantial ongoing complex work, including cross-country, business unit, and project billing, payment scheduling, technical support, and new technology onboarding. One key value proposition for service providers is to handle these heavy-lift tasks that neither original vendors nor clients want to manage independently.

This explains why, even as cloud providers enhance self-service capabilities, partner ecosystems persist long-term. After years of accumulation, Cheetah Mobile has integrated into the partner ecosystems of multiple global major CSPs. As these CSPs continuously add models and AI services, the AI resources Cheetah Mobile can connect to expand accordingly. Meanwhile, as clients move from traditional cloud to Token, multi-model, and Agent-based solutions, its service offerings have opportunities to expand along with customer needs.

Cheetah Mobile's existing base of Chinese companies going global provides an entry point for this expansion. During the mobile internet era, the company served many Chinese companies expanding overseas. As they enter the AI era, their resource needs further extend to overseas cloud, model access, and AI applications. While customer needs have changed, Cheetah Mobile is seeking new ways to deliver value in its relationship with clients.

Of course, whether this ecological position can translate into a real competitive moat requires more data validation, including customer retention, revenue concentration, unit economics, and the share of higher-value services. At this stage, a more accurate judgment is: Cheetah Mobile has established a certain business scale and customer entry point between Chinese companies going global and overseas cloud/AI resources. However, the long-term value of this ecological position still needs subsequent operational data to prove.

III. OpenRouter and Silicon Flow: The Reference Is Not Multiples, But Capital Market Direction

Recently, Stripe announced its acquisition of OpenRouter, providing a symbolic case for the value of AI model routing and management.

OpenRouter connects over 400 models from more than 80 providers, processes over 10 trillion Tokens daily, and serves over 10 million developers and enterprises. In its acquisition announcement, Stripe emphasized that with rapid changes in model quantities and prices, enterprises need dynamic selection based on task complexity, speed, price, and reliability. Media reports suggest the deal value exceeded US$8 billion, though neither side officially disclosed the amount.

The significance of this transaction is not to apply the same valuation multiple to all model aggregation or Token service companies. OpenRouter's global developer ecosystem, product form, growth rate, and strategic buyer premium are all unique. What it truly demonstrates is that the intermediate layer connecting model supply with enterprise demand has entered the strategic asset vision of major tech companies.

Silicon Flow provides another reference point. It has submitted an IPO application to the Hong Kong Stock Exchange, positioning itself as an open, independent Token supply platform. According to HKEX filing materials, Silicon Flow's 2025 revenue was RMB 55.33 million, up over 7x from 2024; its average daily Token throughput reached 578.5 billion in April 2026. However, due to high compute leasing and upfront investments, its overall gross margin was -24% in 2025, remaining clearly unprofitable. Public reports suggest its post-B+ round valuation was approximately RMB 7.74 billion.

OpenRouter and Silicon Flow differ from Cheetah Mobile in business models, technical capabilities, customer bases, and revenue recognition methods. Therefore, Token volumes, revenue scales, or private market valuations cannot be directly mapped onto CMCM. The valuable insight from comparing these three companies is that capital markets are attempting to separately identify the layer of AI model and compute access, inference resource organization, and Token ROI management, rather than only pricing model companies and cloud providers.

For Cheetah Mobile, this provides not a target valuation formula, but a new classification method. As long as its cloud and AI infrastructure business continues to grow and demonstrates independent commercial quality, the market may no longer view it merely as an ancillary revenue stream within traditional global enterprise services.

IV. Why This Change Has Not Been Fully Recognized

A key reason is disclosure granularity.

Cheetah Mobile currently discloses cloud and AI infrastructure revenue separately but does not reveal independent gross margins or operating profits for this business. It remains grouped within the global enterprise services segment, accounted for alongside ad agency operations.

In Q1, global enterprise services achieved an adjusted operating profit of RMB 13.75 million, but this profit came from the entire segment and cannot be directly attributed to the cloud and AI infrastructure business, nor used to infer its independent profit margins. Changes in revenue, costs, and profits from the original ad agency business all affect segment results.

Therefore, at this stage, we can confirm that cloud and AI infrastructure has generated relatively fast-growing revenue and accounts for the majority of global enterprise services. What we cannot yet confirm is the gross margin and operational efficiency of this business itself.

This disclosure ambiguity partly explains market caution, but it also means that if management increases disclosure in the future, or if revenue scale grows large enough to dominate segment performance, conditions for value discovery will significantly improve.

V. Robotics, Internet Profitability, and Cash Provide Second-Tier Support

Cloud and AI infrastructure should be Cheetah Mobile's most noteworthy revaluation variable, but it is not the only change.

In Q1, robotics and other business revenue surged 175.9% year-on-year to RMB 51.2 million, while operating losses narrowed 57.1% year-on-year. Management disclosed that smart personal mobility products began initial deliveries to overseas and domestic industry clients in Q2. The robotics business remains in its investment phase, making it more suitable as a growth option than the primary valuation anchor at this stage. However, simultaneous revenue growth and loss narrowing indicate that commercialization is progressing.

The role of the internet business is also evolving. While traditional online advertising continues to decline, internet value-added services revenue reached RMB 98.3 million in Q1, up 8.2% year-on-year, accounting for 72.8% of internet services revenue. The internet services segment still contributed RMB 15.2 million in adjusted operating profit. It may not be the company's most imaginative future segment, but it continues to provide profits and cash flow to support AI and robotics investments.

The balance sheet offers another layer of cushioning. As of end-March, Cheetah Mobile held approximately US$185.6 million in cash and cash equivalents and about US$100.3 million in long-term investments. A more precise term would be 'cash and investments' rather than 'net cash' (without liability adjustment). These assets should not be simply added dollar-for-dollar to equity value, but they do reduce the company's reliance on external financing during new business investment phases.

Thus, a more complete structure emerges: cloud and AI infrastructure provides the primary revaluation clue; robotics offers longer-term growth options; the internet business continues to contribute legacy profits; cash and long-term investments provide financial cushioning.

VI. What the Q2 Results Really Need to Validate

Cheetah Mobile has confirmed it will release Q2 results on September 11. The market's top priorities should not be whether total revenue slightly beats or misses expectations, but whether revenue mix changes continue.

First, can cloud and AI infrastructure revenue maintain relatively fast growth? Absolute revenue and its continuity matter more than Token usage, as this determines whether demand can truly convert into commercial value.

Second, does the business mix within global enterprise services continue to improve, and can management provide more clues about the cloud and AI infrastructure business's profit model? Before independent disclosure, segment profits should not be directly attributed to this business.

Third, can the robotics business sustain revenue growth while further narrowing operating losses? For a business still in early commercialization, growth quality matters more than mere shipment volumes.

Fourth, can the internet business continue to contribute adjusted operating profit, and does the company's cash and investment scale remain stable? Only if legacy businesses and the balance sheet continue to provide support can new business growth avoid being interrupted by financing pressures.

Risks remain clear: AI infrastructure services may face model price declines, bargaining power from upstream cloud providers, and product commoditization; robotics has yet to achieve segment profitability; declines in traditional advertising could continue to drag on consolidated results. Therefore, revaluation is not inevitable—it requires revenue growth, profit improvement, and more transparent disclosure to drive.

However, the reason Cheetah Mobile now warrants re-examination is not just "the company is also doing AI."

The truly significant transformation lies in the fact that cloud and AI infrastructure, along with robotics—its two emerging business sectors—already account for nearly 40% of its revenue. These segments are poised to become the primary revenue drivers in the upcoming quarters. Should the Q2 results continue to validate this trend, market discourse may pivot from questioning "whether Cheetah Mobile has a compelling AI narrative" to deliberating "to what extent this company should still be valued under the traditional internet framework."

Financial results should not be viewed as mere data points for speculation; rather, they serve as a crucial window to assess whether the current valuation framework requires a fundamental shift.

This article is crafted based on publicly accessible information and is intended solely for industry and company research reference. It does not constitute any form of investment advice.