One of my main learning methods (and hobbies) is reading financial reports. The quarterly financial report disclosures of listed companies, including announcements, press releases, and analyst conference call minutes, provide us with a wealth of financial and business information, which is a treasure trove of knowledge that cannot be ignored. In the past two years, due to the unattractive performance of Hong Kong stocks and Chinese concept stocks, I have spent most of my time reading financial reports of U.S. tech stocks—including the largest companies by market value such as Microsoft, Apple, and NVIDIA, as well as IBM, which has significant influence in certain vertical markets.

When I was in middle school (around the time of the first Nasdaq bubble burst), IBM was one of the tech giants alongside Microsoft, albeit with a lower market value but still a heavyweight player. Over the following two decades, IBM underwent multiple business reorganizations, establishing an enterprise software-centric business model, but also falling out of the ranks of "tech giants" and widening the gap with frontline tech companies like Microsoft.

As of today (September 11, 2024):

IBM's market value is only 1/16th of Microsoft's, or more precisely, 6.14% of it. Last quarter, IBM's revenue was only 1/4th of Microsoft's, and its net income was only 1/12th. Non-GAAP net income was also only 1/10th of Microsoft's. Despite being much smaller in scale, IBM's revenue growth rate for the previous quarter was only 2%, compared to Microsoft's 15%. However, IBM's net income growth rate was slightly faster than Microsoft's (14% vs. 10%).

It's worth noting that the above comparisons are not entirely fair, as IBM completed a spin-off of its IT infrastructure operations business in 2021, which became an independent publicly traded company called Kyndryl. However, even including Kyndryl, IBM's revenue would still only reach about 30% of Microsoft's. Moreover, Kyndryl was spun off due to its thin margins and limited appeal to capital markets, with a current market value of only about $5 billion (equivalent to 1/600th of Microsoft's).

In this generative AI-driven era, Microsoft holds a far superior strategic position compared to IBM: the former is OpenAI's largest external investor, and its Azure cloud is the most commonly used platform for AI training, with GPT fully integrated into software and services like Office, Teams, and Bing. IBM, on the other hand, has become a less significant player, with its former glory represented by IBM Watson now faded, and the company barely making it into the second tier of AI technology. In the foreseeable future, the gap between Microsoft and IBM is far more likely to widen than narrow.

So, the question arises: how did IBM sink to this level? It's worth noting that just twelve years ago, on September 11, 2012, the gap between Microsoft and IBM was negligible:

On that day, IBM's market value was $233.4 billion, while Microsoft's was $256.8 billion, making them roughly equal in size. For that quarter (July-September 2012), IBM's revenue was $24.7 billion, compared to Microsoft's $16 billion; IBM's net income was $3.8 billion, while Microsoft's was $4.5 billion. They were still on par. Over the following twelve years, Microsoft's market value grew nearly 12-fold, while IBM's (taking into account the spin-off) remained stagnant. The widening gap in market values closely mirrors the gap in net income, indicating that this change is fundamentally driven rather than a temporary market sentiment.

Interestingly, both companies have had relatively stable management teams over these twelve years: IBM was led by Ginni Rometty from 2012 to 2020, succeeded by Arvind Krishna in 2021; while Microsoft has been led by Satya Nadella since 2014. Thus, the question can largely be simplified to: What mistakes did IBM under Rometty make compared to Microsoft under Nadella? Or, what did IBM fail to do correctly?

There are undoubtedly many reasons. Without having served as a multinational CEO or having a technical background, I can only offer my outsider perspective. In hindsight, IBM made mistakes in at least three significant directions, in descending order of importance:

Not immediately betting on cloud computing, especially public cloud services, thus failing to adapt to the trend of IT services moving to the cloud; not keeping up with the shift in AI technology towards deep learning, rendering its decades of AI technology accumulation quickly obsolete; and not placing any bets on consumer-facing (To C) businesses, thus losing out on more possibilities (though betting on them might not have guaranteed success either).

Starting with the first point. Over the past two decades, the most significant trend in global IT services has been cloud computing: the shift from companies building their own IT systems to outsourcing public cloud platform services, resulting in the full "cloudification" (outsourcing) of IT infrastructure and even software services. Amazon was the first to venture into this, with AWS becoming its most profitable business (far surpassing its main e-commerce operations). Microsoft followed closely behind. Even before becoming CEO, Nadella was instrumental in Microsoft's transition to cloud computing, facilitating the integration of its database, Windows server, and development tool businesses with the Azure cloud platform. Upon taking over as CEO, Nadella resolutely and unhesitatingly increased investments in Azure, eventually making it Microsoft's revenue growth engine and largest single revenue segment.

Technically and product-wise, cloud computing has limited relevance to Microsoft's existing businesses. Microsoft's strengths in traditional PC and server software do not directly translate to cloud computing services, and the "cloudification" of the former is a long and painful process. The so-called "synergy between Microsoft's traditional and cloud computing businesses" primarily refers to sales synergies—Microsoft's sales system (including direct salespeople and distributors) covers a large number of enterprises, allowing them to recommend Azure cloud services. Microsoft's long-standing customers can also receive discounts when purchasing Azure. This sales-based "advantage" is also shared by IBM and even Oracle, albeit with varying sales coverage.

In short: Microsoft's technological accumulation in the old era does not guarantee that its Azure can catch up with Amazon's thriving AWS in the new era. AWS's predecessor was established as early as 2002 and began offering full services externally in 2006; Azure's predecessor was only established in 2008 and began offering services externally in 2010. In the rapidly evolving cloud computing industry, a 4-6 year gap is significant, requiring double the effort to catch up. When Nadella took charge of Microsoft's cloud business, it was already the final window for a full transition to cloud computing; had he waited another two years, Azure might have fallen behind even newer entrant Google Cloud! Nadella's decision to "go all-in on Azure" was of immense strategic significance that cannot be overstated!

IBM's lag in cloud computing can almost entirely be attributed to a "late start." In 2010, IBM began exploring cloud computing; it wasn't until 2013, after acquiring SoftLayer, that it established a truly cloud-focused service division. However, it wasn't until 2017 that IBM solidified its cloud strategy centered on "hybrid clouds" (a combination of public and private clouds), reinforced in 2018 with the acquisition of Red Hat. By then, Amazon AWS, Microsoft Azure, and Google GCP had already established themselves as the top three players, leaving IBM with a narrow market niche primarily among large enterprise customers preferring hybrid clouds.

Ironically, according to IBM's official narrative, Rometty's greatest achievement as CEO was "establishing IBM's hybrid cloud strategy"—which is true, except that this strategy should have been established in 2012-2013, not 2017-2018! "Justice delayed is justice denied," and similarly, a delayed correct decision can only become mediocre. Moreover, hybrid clouds are not an IBM-exclusive technology; Amazon, Microsoft, Google, and even Oracle can all offer them. Even within this narrow niche market, IBM's position is far from secure.

Moving on to the second point. IBM once maintained a fifty-year lead in AI technology. Friends born in the 1970s and 1980s will likely remember the 1996 match where "Deep Blue" defeated Garry Kasparov. Americans may also recall IBM Watson's impressive victory over human contestants in the Jeopardy! knowledge competition in 2011. A key strategy during Rometty's tenure as CEO was to commercialize IBM's AI technology through Watson solutions, with a primary focus on the healthcare industry.

It turned out that the European and American healthcare industries were too complex, involving numerous regulatory and ethical issues, making them less suitable for AI transformation at the time. Watson's commercial performance was initially promising but declined. More importantly, an AI technology revolution occurred in 2012-2013, when deep learning based on neural networks not only replaced traditional knowledge graphs (symbolism) but also surpassed statistical learning and other traditional machine learning techniques, becoming the most efficient and widely used foundational AI technology. In just a decade, deep learning revolutionized internet content distribution and advertising delivery systems, paved the way for emerging industries like autonomous driving and large language models (LLMs), and became the dominant force in academia.

As a long-standing AI technology leader, IBM failed to keep up with the times. Whether due to management decision errors, insufficient resource investments, or inefficient execution is beside the point. What matters is that while Google was acquiring deep learning-based AI startups at a rate of ten per year and recruiting top AI scientists like Ilya Sutskever and Andrew Ng at exorbitant prices, IBM's actions were negligible. As a result, within just 2-3 years, Google usurped IBM's position as the AI technology leader and rapidly applied AI to services like search engines and translation, setting in motion the "technology-application-commercialization" flywheel. Poor IBM didn't admit Watson's failure until 2021, by which time its foundational AI research capabilities had already lagged far behind Google, Amazon, and Meta.

Strictly speaking, Microsoft was also a latecomer in this game, as deep learning has never been its forte. However, Microsoft made a highly correct decision by investing in OpenAI in 2019 and integrating its services into the Azure cloud platform. After ChatGPT's emergence, everyone had to admit that this was Microsoft's most important and successful strategic investment in its history. Its deep collaboration with GPT not only boosted sales of software services like Microsoft Office and Teams but also established Azure as the "premier AI services cloud platform." Since 2023, AI demand has increased Azure's revenue growth rate by at least 5 percentage points each quarter. Now it's Amazon's turn to feel the heat and scramble for countermeasures!

Of course, Microsoft didn't pin all its hopes on external investments; it has been continuously investing in internal generative AI research and development. For instance, it co-developed the trillion-parameter Megatron large model with NVIDIA and is still developing and iterating on large models. Over the past five years, as Azure gradually unleashed significant profitability and cash flow, Microsoft has been able to allocate more substantial resources to foundational research directions like generative AI, enabling its mature businesses to "nourish" emerging ones. IBM or Oracle, on the other hand, don't have such abundant resources to squander. Success breeds success, just as money begets money; the key lies in proper resource allocation.

Finally, the third point. Since selling its PC business to Lenovo in 2005, IBM has had virtually no significant consumer-facing businesses. For nearly two decades since then, IBM has shown no interest in consumer-facing businesses, be it consumer internet, consumer hardware, or content services. Fairly speaking, this isn't necessarily a "mistake" per se, as IBM simply doesn't possess the consumer-facing business gene. Even if it had bet on some consumer business over the past decade, it's hard to predict how successful it would have been.

The problem is that Microsoft, at the same time, didn't possess much of a consumer-facing business gene either, yet it persisted, failing repeatedly but fighting on bravely. Its investments in gaming spanned the tenures of three CEOs: Bill Gates, Steve Ballmer, and Satya Nadella. While its investments in smart hardware were disastrous in smartphones, it achieved some success in tablets, maintaining Microsoft's strategic presence. Its investments in consumer internet, primarily Bing search engine and LinkedIn, have been generally successful, especially after integrating with generative AI, enhancing their strategic value. Incidentally, Bing's launch and LinkedIn's acquisition both occurred during Nadella's tenure as CEO.

It's evident that during Ballmer's tenure, Microsoft was indeed inept at consumer-facing businesses, often flip-flopping and lacking direction, with acquired quality businesses often floundering. However, under Nadella, Microsoft's consumer-facing businesses matured, with at least mediocre performance. Financially, all of Microsoft's consumer-facing businesses are now profitable, even its historically money-burning gaming division (thanks to a shift in Xbox hardware platform positioning). From a capital market perspective, they are no longer dragging down Microsoft's market value but instead contributing positively to it in increasingly significant ways.

Nadella's attitude towards consumer-facing businesses is encapsulated in his explanation for the Activision Blizzard acquisition in early 2022: Microsoft cannot afford to be absent from gaming, a massive consumer segment with over 3 billion users. By the same token, one can infer that Nadella believes Microsoft must venture beyond its "comfort zone" of enterprise-focused businesses to maintain its status as a tech giant. This is not only to establish direct connections with consumers and cultivate user loyalty but also to create business synergies—such as the synergy between AI and Bing, and between Azure and cloud gaming. In contrast, none of IBM's CEOs over the past two decades have made similar judgments; they all believed that IBM could maintain its status as a tech giant solely through enterprise-focused, and even profitable large enterprise, businesses. History has proven them wrong.

However, since IBM has made more unforgivable mistakes in the two strategic directions of cloud computing and AI, the mistake of not placing any bets on the consumer business is not as important and can even be ignored. Ironically, the most glorious moment in IBM's history was precisely when it was the strongest in the consumer market - from the 1980s to the early 1990s, IBM PCs led the first wave of the information technology revolution into millions of households, until Compaq, HP, Dell, and other vibrant new manufacturers came from behind. In the consumer computer market, Apple was once defeated by IBM, but within just a few decades, it stood up again and became a technology giant fully based on the consumer market. Things change all the time, but it depends on people. Whether it's "genes" or "historical accumulation," ultimately, it relies on people to execute.

Therefore, we can better understand why CEOs in American listed companies can always receive extremely generous compensation: According to Bloomberg News statistics, the average compensation of CEOs of American listed companies in 2022 was 400 times the average level of employees! Star CEOs like Elon Musk can take away compensation packages worth billions of dollars each year. Nadella's 2023 compensation package was $48.5 million; in 2013, before he became CEO, as Microsoft's senior vice president and head of cloud computing, his compensation package was only $7.6 million. Even considering inflation factors in the last decade, the gap is still enormous!

Is this situation reasonable? Considering the crucial position of the CEO, it is obviously reasonable. During Nadella's ten years as Microsoft's CEO, Microsoft's stock price has risen tenfold; while during Ballmer's previous tenure, it was basically zero growth. If Rometty had made at least one correct decision on the two strategic issues of cloud computing and AI during her tenure as IBM's CEO and implemented it, IBM might still be standing among technology giants today, with a market value of trillions of dollars instead of $185 billion. The American corporate governance structure gives CEOs almost unlimited business decision-making power, so CEOs should take responsibility for all mistakes and be rewarded for all achievements. This is the so-called "unity of power and obligation."

From this, we can further infer that in Chinese companies adopting the American corporate governance structure, mainly Chinese internet companies, CEOs have even greater power and responsibility. They not only enjoy the institutional power granted by American companies but also possess the unique non-institutional power of the Chinese human society, allowing them to implement their will more efficiently and thoroughly. When they make correct or wrong choices, their impact on the company is even greater. Therefore, it is perhaps somewhat extreme but still reasonable to say that at least half of the success or failure of internet giants can be attributed to the CEO.

Regarding the above question, I would like to expand on it further, but that should be the theme of another article.

This article did not receive any funding or endorsement from Microsoft or IBM.

At the time of publication, the author of this article does not directly hold shares in Microsoft or IBM but may hold them indirectly through funds or trust plans.