"Money-guzzling Behemoth" Starts to Give Back: How Do Listed Tech Companies Make Money from AI?

Three years after ChatGPT ignited the global tech industry, massive AI investments have finally begun to yield scalable returns.

AI-driven profits from Chinese and U.S. tech giants have exceeded expectations. Google and Meta saw their Q2 net profits surge by 19.4% and 36% year-over-year, respectively; Alibaba Cloud's revenue surpassed 63.5 billion yuan, nearly matching the growth rate of Microsoft's Intelligent Cloud revenue in Q2; Tencent also reported rapid growth in AI-related revenue.

However, amidst this high growth, AI darlings like Figma and C3.ai reported disappointing earnings, triggering sharp stock declines and signaling a shift in AI investment logic.

Capital focus is shifting from "input" to "output." What truly matters is no longer how much is invested but the marginal value generated, unit economics improved, and whether these capabilities can be monetized at scale.

So, among Chinese and U.S. tech companies, who will truly crack the business model of AI and emerge as the next-phase winners?

AI Enters the Payoff Phase: Clear Signals from Giant Earnings Reports

Over the past three years, AI investments in the tech industry were seen as a "must-answer question," but it wasn't until the 2025 interim reports that AI clearly emerged as a powerful engine driving both revenue and profit growth for tech companies.

Tech giants on both sides of the Pacific delivered results that exceeded expectations.

In the U.S., Alphabet (Google's parent) reported Q2 revenue growth of nearly 14% year-over-year, with net profit growing even faster at close to 20%. This performance was supported by over $22 billion in capital expenditures, mostly directed toward computing power and data centers.

Microsoft's Intelligent Cloud revenue grew by over 25%, boosting overall profits; the company expects Q3 capital expenditures to exceed $30 billion, far surpassing market expectations. This indicates a focus not just on short-term financial performance but also on building infrastructure for future AI expansion.

Meta leveraged AI in ad recommendations and content ecosystems to achieve 22% revenue growth and 36% profit growth, with plans for additional billions in spending to address computing bottlenecks.

Source: Smart Little Titan

On the domestic front, Alibaba's Cloud Intelligence Group reported revenue exceeding 63 billion yuan in its interim report, becoming a growth pillar. Its e-commerce and local services businesses also saw efficiency gains from AI applications. Tencent, by embedding AI capabilities into gaming, social, and advertising scenarios, achieved revenue of over 180 billion yuan, further solidifying its scale advantage.

Like their U.S. counterparts, Chinese companies demonstrated that AI is becoming a foundational capability across their ecosystems.

Source: Tencent

Behind this earnings surge is the fact that, after years of trial and error and heavy investment, tech companies have finally completed the initial loop from model development and computing infrastructure to real-world applications.

Massive capital expenditures are no longer about "burning money for the future" but about sustained investments with clear returns, preparing for the next phase of competition.

Investor attitudes have also shifted, from concerns about uncontrollable costs to recognition of the revenue and profit improvements AI brings. As the head of AllianceBernstein noted, while capital expenditures still seem "endless," investors' attitudes have changed dramatically because these tech giants' AI investments are now showing returns. The market's positive reaction, driven by growth in cloud computing revenue and AI services sales, is "truly remarkable."

All this indicates that the AI industry has entered a new phase where investment and output reinforce each other in a virtuous cycle. Moving forward, demonstrating "how AI can make money" will be a must-answer question for tech companies.

From "Spending More" to "Using Better": Three AI Industry Paths Emerge

The flip side of the interim reports is a shift in market standards. AI investment is no longer a bonus; investors now care more about who can truly convert technology into revenue and profit. The capital market's patience for AI transformation is wearing thin, as evidenced by the earnings "bombs" and stock plunges of U.S. AI darlings Figma and C3.ai.

On September 4, Figma's stock plummeted over 21% intraday after its Q2 revenue growth of 41% fell short of expectations, prompting investor concerns about valuation support. Similarly, U.S. AI star C3.ai's stock dropped over 11% due to poor performance and worrisome financial metrics.

This reflects market concerns about AI disrupting traditional application ecosystems. Clearly, grand narratives about AI technology alone are no longer enough to retain market patience. From the first half of the year, three distinct AI application paths have emerged.

First, tech giants have chosen the path of "heavy infrastructure and ecosystem building."

Alibaba invested about 50 billion yuan in computing power and model R&D in the first half, accounting for over 10% of revenue. Cumulative investments over the past four quarters exceeded 100 billion yuan, aimed at building full-stack capabilities from underlying computing power to large models and industry solutions. Tencent followed suit, with its 22.9 billion yuan AI investment slightly smaller but more focused on deep integration with core businesses like social, gaming, and financial payments to enhance monetization efficiency. A 20% year-over-year growth in Q2 marketing services revenue is a direct result.

Image: Tencent Yuanbao x Fearless Contract Mobile Game

Clearly, tech giants are channeling funds into AI infrastructure and business restructuring.

In contrast, vertical players are adopting a "strong scenario" strategy.

These companies don't pursue general capabilities but concentrate resources on specific areas to build absolute advantages. For example, SenseTime's 2.12 billion yuan investment, accounting for 89.8% of revenue, delivered stellar AIGC revenue of 1.816 billion yuan, or 77% of total revenue, proving the viability of a specialized, high-investment path. Other examples include iFlytek's 3.5 billion yuan revenue in smart education, Kuaishou's AI-driven content ecosystem contributing billions in revenue, and Meitu's over 30% net profit growth in AI imaging and design.

Source: iFlytek Smart Education

An exception is Baidu, whose strategy straddles both approaches, maintaining investments in foundational models while strengthening its "AI-native" strategy through scenarios like autonomous driving and smart cloud.

The third path treats AI as a product upgrade and business model extension.

Traditional software and internet companies showcased a pragmatic shift toward "rapid transformation and conversion." While they may lack top-tier model capabilities, their strengths lie in mature business scenarios and user bases. For them, AI is the best tool for product renewal, boosting payments, and extending lifecycles.

For instance, Kingsoft Office launched new features like AI document editing, Lingxi Voice Assistant, WPS AI PPT, and WPS Knowledge Base based on WPS AI 3.0, driving a 14% revenue growth in office services. 360 Company explored new growth points through "AI + search" and "AI + security" initiatives. These companies successfully "activated old businesses" and won market recognition.

Source: Kingsoft Office

Overall, giants' ecosystem building, vertical players' scenario breakthroughs, and traditional companies' product renewals are all answering the same question: Has the AI business model truly been cracked?

The emergence of these three paths reveals an industry trend: Successful AI strategies must move beyond vague future promises and focus on precision and pragmatism.

Capital is no longer buying mere "AI stories." Tech companies must translate technical prowess into solid financial performance or clear growth potential to retain investor confidence in this late-stage race.

From Exploration to Growth: A New Valuation Cycle Arrives

If we divide the AI industry's lifecycle into stages, it has likely transitioned from infancy to "growth." The focus has shifted from infrastructure and model training (the "investment phase") to commercialization and efficiency (the "return phase").

Looking ahead, the industry will enter a phase of elimination and reshaping. Companies that achieve valuation leaps will either significantly optimize costs and improve efficiency through AI or create new AI revenue streams—or both.

From this perspective, tech giants' long-term growth remains relatively certain, supported by a triple guarantee: "AI-ization of core businesses + AI-led emerging businesses + strategic investments in 100 major AI applications."

Strategic investments in "100 major AI apps" can be seen as version 2.0 of the old "traffic and capital empowerment." For example, Alibaba owns large-model products like Tongyi Qianwen while also investing in nine companies, including SenseTime and Zhipu AI, covering the entire value chain from computing power to applications. Tencent and Baidu follow a similar playbook, developing their own AI applications (e.g., Tencent Yuanbao, Wenxin Yiyan) while investing in companies like Kuaishou and Baichuan Intelligence to strengthen their positions in content ecosystems and model capabilities.

Source: The Paper's Alignment Lab, "2025 Global Top 100 AI Applications"

With AI as the valuation core, tech giants' internal and external empowerment could benefit from AI application development, likely leading to systemic revaluation.

Meanwhile, leaders in vertical AI application sectors may also see valuation resets. This earnings season shows that, despite competition from giants, these players have carved out high-growth paths through extreme scenario focus and model innovation. For instance, Meitu and Kunlun Tech reported significant year-over-year AI revenue growth, with substantial user bases and revenue from overseas, proving the global potential of vertical AI applications.

Source: Guohai Securities

In the future, more industries will enter the "AI 2.0" phase. Investment focus will shift from a few tech leaders to traditional industry frontrunners driving AI adoption—such as agriculture, manufacturing, and finance—using AI to optimize production, services, and efficiency.

Early wins don't guarantee final victory. Many of today's fluctuating AI stars may not emerge as ultimate winners, as new players with industry expertise could disrupt the landscape, much like the "Internet+" era upended traditional sectors.

Moreover, thanks to the digital and informational foundations of the internet era, AI industry competition is evolving faster. Many companies will accelerate their runs on their respective tracks. For investors, the most robust approach is no longer identifying who has AI technology but judging who can already convert AI into tangible financial results.

In this sprint of evolution, a new valuation logic is taking shape. Who will be the next AI dark horse?

Source: Hong Kong Stock Research Society