I frequently peruse the financial reports of technology companies from both China and Silicon Valley. This is not merely a professional obligation but also a personal passion. Reality often surpasses science fiction in its intrigue, and delving into the progression of human technology and society through genuine financial reports is both captivating and rewarding. While financial figures are crucial, the long-term, underlying logic behind them is even more so: What are the development paths of tech giants? What fuels their growth? And how do their leadership teams envision and respond to the future?

Over the past three years, I have read nearly every quarterly report of all tech companies globally with a market value exceeding $100 billion. My most profound takeaway is that the era of 'explosive growth' for tech giants, characterized by double-digit or even triple-digit growth rates, has irrevocably come to an end. Whether it's Microsoft, Amazon, Google in the US, or Alibaba, Tencent, Baidu in China, the direct impact of generative AI on their performance has been gradual and limited, with revenue growth rates in the teens becoming the norm. Companies like Apple have even witnessed single-digit income growth for multiple quarters in succession. This stands in stark contrast to the mobile internet traffic dividend era over a decade ago, when tech giants experienced explosive, all-encompassing, and seemingly inexhaustible growth. Nowadays, only companies in the computing power chain, such as NVIDIA, can enjoy such explosive growth, and we all recognize that the computing power chain constitutes just a small segment of the tech industry.

Do not misunderstand; this is not because the AI revolution is a misnomer or because the momentum for human technological progress has waned. The correct interpretation is that the revenue base of tech giants is already immense, and the level of informatization in human society is already high. Basic arithmetic principles dictate that the performance of these giants will inevitably 'revert to the mean.' For many years, the tech industry, particularly the internet sector, has been engaged in a series of 'sprints': Whenever a new trend emerges, everyone rushes in, striving to secure their territory as swiftly as possible, then waiting for the next trend, and so on, in a cyclical pattern. This development paradigm has reached its conclusion – what lies ahead will be true 'marathons.' As we all know, a marathon is not a mere collection of sprints!

Recently, I read Ant Group's 2025 annual sustainability (ESG) report, and two points caught my attention: First, Ant reaffirmed its 'AI-first' strategy, proposing 'dual value creation towards the AI era.' Second, it suggested that AI introduces new possibilities for inclusive services, with traditional 'digital services' evolving towards more intelligent 'digital-intelligence services.' Almost everyone concurs that AI will be the fundamental driving force for the global tech industry in the next decade and beyond. The pivotal question is, how precisely should AI applications be realized? Or, to phrase it differently, how can we ensure that AI creates both commercial and social value? After all, when Sam Altman founded OpenAI, he posed a celebrated question: 'Will AI destroy humanity?' This question remains unanswered, and the November 2023 'palace coup' at OpenAI is intimately connected to it, a memorable event.

Applying AI technology to provide inclusive services to all users, even ascending to the level of 'digital-intelligence inclusivity,' is a prudent approach. Some may dismiss this statement as 'grandiose and empty,' but a historical review reveals that the growth of tech giants has always been rooted in expanding user bases and lowering service thresholds. The PC revolution of the 1990s lowered the threshold for computers, the internet revolution of the 1990s-2000s lowered the threshold for information dissemination, and the smartphone and mobile internet revolution of the 2010s further lowered the threshold for all personal computing and communication services: These are all inclusive. Microsoft provides inclusive operating systems, Amazon and Alibaba provide inclusive e-commerce services, Meta and Tencent provide inclusive social services, and they have all become companies worth tens or even hundreds of billions of dollars. Can we claim that inclusive AI services are solely provided by large model vendors like OpenAI? No, that's not comprehensive – large models are merely a foundation, and chatbots are just one starting point for AI applications. All internet services, and even all mature service industries, are destined to be reimagined in an AI manner.

About two months ago, I read an interview with Ant Group's management team, and one sentence left a profound impression on me: 'Having money to spend and having the life to spend it are eternal themes.' I certainly comprehend what that means! Over the past two years, I've lost ten kilograms, basically quit drinking (except when traveling), adopted a low-carbon diet, and developed a habit of outdoor exercise. The so-called 'having money to spend' doesn't imply becoming an overnight millionaire but rather having ample and satisfactory financial resources steadily over time. The so-called 'having the life to spend it' refers to living a high-quality life for as long as possible. In the past era of explosive growth, we disapproved of these concepts and even ridiculed them; what was fashionable was 'trading life for money' and 'pushing oneself to the limit to achieve financial freedom.' Most of my friends work in the financial and internet industries, and in recent years, they should have increasingly realized the absurdity of 'trading life for money,' especially after seeing their medical examination results and local real estate prices.

Let's elevate this to a philosophical level: 'Trading life for money' corresponds to the logic of 'explosive growth' or the 'era of windfall profits,' whose core idea is to seize opportunities for a big score, recognizing only short-term high-growth opportunities without considering sustainable development. 'Having money to spend and having the life to spend it' corresponds to the logic of the 'era of compounding returns,' which is the era we are currently in.

The power of compounding returns far surpasses the intuitive impression of most people. An annualized growth rate of 12% means doubling in six years; an annualized growth rate of 8% means doubling in nine years. Warren Buffett's annualized return rate is around 20%, which has been sufficient to make him the most successful investor in the world.

Only robust companies can fully benefit from compounding returns. As a former investment analyst, I recall that when modeling popular companies for the capital market, we often predicted dramatic short-term growth of 30-50%. Many companies indeed achieved this, but then stagnated or even experienced negative growth, eventually fading from mainstream view.

'Having money to spend,' or wealth preservation and appreciation, can rely on windfall profits over a very short timeframe but ultimately depends on compounding returns. 'Having the life to spend it,' or health management, is a thorough compounding problem: As far as I know, there seems to be no magic pill in the world that can make someone healthy overnight.

Intriguingly, finance and healthcare are precisely the two industries with the fastest progress and most novel developments in AI transformation, and they are also the most likely to realize 'inclusive services' driven by AI. Let's commence with finance: In 2024, Goldman Sachs explicitly mentioned in its earnings call that a quarter of its employees are now 'technology developers.' Just this June, Goldman Sachs' Chief Information Officer announced to all employees the launch of 'GS AI,' an artificial intelligence assistant. This is not a general application but a tailored solution for the diverse needs of business departments such as investment banking, research, asset management, wealth management, and more. Moreover, Goldman Sachs is not even the first Wall Street investment bank to promote an AI assistant at the group level; Citibank, Morgan Stanley, and Bank of America have already made similar attempts. Many people's understanding of 'AI applications in finance' still halts at customer service robots, but in fact, AI's involvement in financial business processes is nearly all-encompassing. In another year or two, we may not find any financial business segment that hasn't been deeply transformed by AI!

Wealth management is the fastest-evolving sub-sector in finance: Wealth management departments of UBS, Morgan Stanley, and JPMorgan Chase have introduced AI platforms based on machine learning and natural language recognition, analyzing market data and customer trading behavior to provide real-time investment advice to clients. 'Robo-advisor' companies represented by Betterment, Wealthfront, and Pefin have amassed a substantial customer base, with Wealthfront's assets under management reaching $80 billion (as of March 2025). AI-driven wealth management is one of the hottest trends in the Silicon Valley venture capital circle in recent years. According to Forbes magazine's estimates, in the US alone, AI investment advisors may manage over $400 billion in assets.

In China, a batch of AI investment advisors or even more comprehensive 'wealth managers' have also emerged, such as Ant's 'Maixiaocai,' JD.com's 'Jingxiaobei,' and Yingmi Fund's 'AI Xiaogu.' Just a few days ago, a friend working in family trust services told me that he believes AI wealth management will become a great 'equalizer,' enabling ordinary people to enjoy financial services traditionally accessible only to ultra-high-net-worth individuals at a lower cost – customized asset allocation plans based on family circumstances, specific product recommendations, regular exclusive evaluation reports, real-time advice for market emergencies, tax planning, and even wealth succession plans... These services, which traditionally required enormous asset sizes and high management fees or commissions, can now be efficiently generated by AI, and their accuracy may even surpass that of humans. This is the power of 'inclusive finance'!

Incidentally, I believe that the financial industry may be one of the very few industries that truly necessitate 'vertical large models' due to its complexity and involvement in sensitive areas such as customer privacy, information security, risk control, and regulation. Ant and Duer are both developing their own vertical large models; even general large models need to be fine-tuned for financial scenarios, and many financial institutions are doing this with open-source large models like DeepSeek. In the era of generative AI, the 'technological attribute' of finance will become more important than ever, and the development of 'inclusive finance' will largely hinge on the progress of core technologies such as foundation large models. Financial institutions must become tech companies, not just pay lip service!

Now let's discuss healthcare. As early as around 2010, IBM Watson, as one of the first professionally commercialized AI assistants, attempted to break into healthcare. Unfortunately, AI technology based on knowledge graphs couldn't handle such a demanding task at that time. Times have changed, and AI based on deep learning technology and capable of natural semantic recognition can finally transform the healthcare industry. Earlier this year, the US Department of Health and Human Services published a 200-page 'AI Healthcare Strategic Plan,' proposing to apply AI to the 'five fundamental areas' of healthcare: medical research, new drug development, medical services, humanitarian services, and public health. In short, it aims to reshape the entire healthcare industry chain with AI technology. In fact, China is also swiftly leveraging AI to transform healthcare, and there's even a possibility that it will surpass others: Within just two or three weeks after the launch of DeepSeek R1, at least 92 public hospitals connected to DeepSeek. Large procurement orders for artificial intelligence systems worth tens of millions of yuan have emerged one after another in hospitals across the country, and perhaps the nearest tertiary hospital to your home has just completed such a procurement.

The significance of AI in medical and pharmaceutical research and hospital internal management has long been recognized, but I believe what's more crucial is AI-based 'medical service equalization': letting AI reduce the difficulty and complexity for ordinary people to access medical services while minimizing ineffective labor in the public health system. Finding a doctor, measuring health, reading reports, accompanying visits, inquiring about medical insurance... all these require considerable time, and money doesn't always help, let alone without it. An important selling point of many high-end credit cards in China is 'several accompanying visit services per year.' I still remember the painful experience of over three years ago when I developed palpitations after having my wisdom teeth extracted and feared I had heart disease – it took me a whole day to study and prepare just to determine which department I should register with, and another half-day running up and down for examinations during the visit. Fortunately, the examination results were normal, which made me not dwell on the difficulties I experienced, but that doesn't mean the difficulties didn't exist.

Now, more than one domestic tech giant has introduced AI medical service tools for individual users, such as ByteDance's Xiaohe AI Doctor and Tencent's AI Health Management Assistant. About five days ago, Ant just launched the AI health app 'AQ,' encompassing over a hundred functions such as health science popularization, consultation for visits, report interpretation, and health records. There are two features of Ant AQ that appeal to me the most: The first is the personal health record, which integrates information on medical treatment, medication, exercise, and diet, coupled with the integration of various wearable smart hardware, to establish a comprehensive and personalized 'health database,' providing timely health suggestions to help users be prepared. The second is 'Cloud Visit Accompaniment,' which currently covers more than 200 hospitals nationwide. When users go to a hospital for offline treatment, AI can provide online guidance, helping patients efficiently complete the entire visit process without detours. These are just two simple examples of AI-driven 'inclusive healthcare.'

I reiterate my point – AI in healthcare is a formidable 'equalizer,' and its full potential remains largely untapped. For instance, the aforementioned Ant AQ has already created 'AI avatars' of approximately 200 renowned doctors. Even an elderly individual in a remote or impoverished area can now consult with elite doctors from prestigious Beijing hospitals at any moment. Of course, it's essential to acknowledge that the current capabilities of 'AI avatars' are somewhat limited, with a narrow scope and inability to replace actual medical treatment. But what might the scenario be in another five or ten years? Since ChatGPT's inception just over two years ago, we have grown accustomed to leveraging AI for information retrieval, article writing, and programming, significantly boosting our work efficiency. Although AI may never fully replace real doctors, if it can merely assist them in some daily tasks, it could translate to improved quality of life for thousands, or even millions, of ordinary people. The societal significance of such technological advancements cannot be overstated! I fervently hope that by the time I and the readers of this article age, such applications have reached full maturity.

Moreover, I wish to emphasize that the exhilarating visions outlined above, be it 'inclusive finance' or 'inclusive healthcare,' will take time to materialize. Their realization hinges not only on technological progress but also on addressing compliance and ethical issues. Regarding their impact on the performance of tech giants, the timeline is even more uncertain. Tech giants must prioritize providing inclusive services to users before enhancing their commercialization efforts. Investors and professional media accustomed to the 'turbulent' growth of the mobile internet era may perceive this development as too sluggish, dismissing it as mere slogans devoid of practicality. Among my acquaintances, many remain skeptical or negative about the entire AI wave, with their core argument being that 'the growth spurred by AI is not as instantaneous as that of the internet in the past.'

As I pen this article, I have just departed from Azerbaijan, the sole oil-exporting nation among the three Caucasus countries. Over a decade ago, it witnessed an 'economic miracle': From 2005 to 2008, due to skyrocketing global oil prices, Azerbaijan's GDP surged 2.7 times, and its per capita income leaped from 15% to 30% of that of the United States! However, this golden period was fleeting. With the decline in oil prices and geopolitical instability, the country's economic growth rate reverted to single digits after 2009, with several instances of negative real GDP growth. Azerbaijan's industrial chain remains incomplete, and it is also a typical developing country in cultural and social terms; last year, its per capita GDP was precisely half of China's. Over half of the ride-hailing cars I hailed on the streets of Baku were imported from China (especially from BYD)!

This serves as an exemplary real-world illustration of the contrast between the "era of quick profits" and the "era of compound interest": Relying on the export of natural resources like oil can propel short-term growth instantaneously, but long-term success pivots on infrastructure, cultural education, scientific and technological research and development, and industrial chain development. Both Azerbaijan and Georgia, where I currently reside, are undertaking large-scale infrastructure upgrades (with numerous projects involving Chinese enterprises) and enhancing public education, indicating their long-standing awareness of this fact. The same principle applies to enterprises, particularly so-called large companies. If you peruse the annual reports of major domestic internet companies, you'll find that almost all of them emphasize similar themes: driving long-term value through technological innovation and pursuing sustainable development. I would add that the crux of sustainable development lies in grasping the essence of each era. In layman's terms, the theme of our era might be "having the means to spend and the longevity to enjoy it"; although some may prefer a more eloquent or theoretical expression.

I adore "Star Trek," a classic science fiction franchise whose most famous line is: "Live long and prosper." When "Star Trek" was conceived, AI was but an ethereal specter, so this line was a mere blessing without specifying how to achieve it. Today, we might append the second half: "With the assistance of AI." Whoever can first turn this phrase into a reality will occupy the pinnacle of the new era, whether it be a major internet company, a startup, or an unknown passerby.

This is precisely why I enjoy delving into the financial reports of large companies: They provide insights into who recognizes the future, who is adequately prepared for it, and who can "realize the future" most efficiently. These reports may not offer complete answers, but they at least provide partial responses that can spark profound reflections.