Four expeditions about dignity and survival

September 3, 2013, Helsinki.

At the press conference where Nokia sold its mobile phone business to Microsoft, CEO Stephen Elop finished his last sentence and then paused for a long time.

He said: We didn't do anything wrong, but somehow, we lost.

Then he choked with sobs and shed tears. Many people in the audience also cried.

At that moment, the entire industry knew that an era had ended.

Nokia, the king of mobile phones, once had a market share of nearly 40% at its peak, but overnight, it lost its territory to the mobile wave.

No one thought that twelve years later, for every 5G mobile phone sold globally, a patent fee would have to be paid to this "dead company". This includes Apple, Samsung, Xiaomi, and also Huawei.

In the same year, in California, Intel was anxious about its absence in the mobile era. In 1999, its market value exceeded $500 billion, and the "Intel Inside" label was everywhere in the industry.

But in the smartphone race, it was absent throughout.

No one foresaw that an even greater crisis was yet to come. Seven years later (around 2020), the era of AI training arrived. NVIDIA's GPU became the most core computing power unit, and Intel's CPU was completely reduced to a supporting role and spectator.

In 2024, its stock price dropped by nearly 60% throughout the year and was kicked out of the Dow Jones Industrial Average. It had held this honor for 25 years, and was finally replaced by NVIDIA.

That day was the most humiliating moment in Intel's history.

In 2020, Huawei experienced its own moment of crisis. The US sanctions were escalated, TSMC cut off its cooperation, the foundry channel for Kirin chips was blocked, and its market share in the mobile phone market dropped from the world's first to the fifth.

It can be described in four words: out of ammunition and supplies.

The outside world's judgment was almost unanimous: Huawei's mobile phone business was over.

Even earlier, there was a company whose "death" was even more silent - Oracle, the world's largest database software company. After the rise of AWS, it was regarded as a legacy company.

Around 2015, almost all technology analysts were writing about the same thing: in the era of cloud computing, there was no place for Oracle.

Four companies. Four endings.

The outside world's judgments about them were later severely slapped in the face by history.

I. Dislocation and Break

Putting these four companies together, we can find that such "deaths" are by no means accidental. They are on the same industrial map, distributed on two different main axes.

One is the computing power main axis, including Intel and Oracle. One is about underlying chips, and the other is about cloud databases and infrastructure. Together, they form the foundation of computing in the PC and early Internet eras. Intel is responsible for making the machines run, and Oracle is responsible for providing a place for data to be stored.

The other is the connection main axis, including Nokia and Huawei. One connects people around the world with mobile phones, and the other connects devices around the world with 5G. In the most glorious era of mobile communications, these two companies were the leaders in the entire industry.

Then, these two main axes simultaneously encountered the same thing, a larger - scale paradigm shift.

The mobile Internet brought down Nokia, the US sanctions brought down Huawei's mobile phone business, NVIDIA and GPUs redefined computing power and kicked Intel out of its home court, and the cloud - native wave made Oracle seem like an old man from the previous era.

Four companies, two main axes, and a nearly cruel systematic judgment.

No one thought at that time that this was not the end.

II. The River Rages On, and the World Didn't Wait for Them

During the years when these four companies were silent, history did not pause; instead, it accelerated.

In 2013, the mobile Internet entered an explosive period. While technology was evolving greatly, dozens of historical events overlapped with each other, expanding like a pressure field in every direction simultaneously.

The number of WeChat registered users exceeded 600 million, AWS's annual revenue exceeded $3.1 billion, Uber expanded simultaneously in 22 countries and more than 60 cities around the world and burned a large amount of money. Data began to explode on a scale that no one could have predicted, and cloud services began to engulf all industries that could be digitized.

What did this mean for Nokia? It had just sold its mobile phone business, and the era it was best at had become a huge tombstone.

For Oracle, all the world's data was flowing to the cloud, and AWS and Azure were turning data storage into a public service like water and electricity.

But an even greater impact was yet to come.

In March 2016, AlphaGo defeated Lee Sedol 4:1. The shock of this event was not about the outcome of a single game, but that it proved to the world that deep learning could really overcome the most difficult cognitive challenges for humans. A breakthrough that was originally thought to be ten years away came earlier.

From that moment on, everyone began to think about the same question: What does AI need?

The answer is: computing power, computing power, and still computing power.

NVIDIA's GPU transformed from a game chip into the hottest commodity in the AI era. In 2019, NVIDIA's market value was about $110 billion, less than half of Intel's.

No one knew that five years later, its market value would exceed $3 trillion and reach the top of the global market value list. Even less could they imagine that in 2025, it would break through $4 trillion and become the first company in human business history to reach this milestone.

Intel stood by and watched all this happen, but just couldn't squeeze into the narrow door to the future.

In 2020, the US escalated its sanctions against Huawei. This became a watershed in the entire technology industry, clearly indicating for the first time that technology had become a core piece in the great - power game.

Huawei's high - end mobile phone business basically stagnated, but it forced the entire Chinese semiconductor industry to seriously think about a question that no one had thought about before.

If the supply chain is cut off, can we make it ourselves?

On November 30, 2022, ChatGPT was officially launched. Within two months, its monthly active users exceeded 100 million, setting the fastest - ever record for the popularization of a consumer - level application. The whole world suddenly realized in the same week that AI was not the future, but the present.

From that moment on, every company began to ask: What is my AI strategy?

And the answer to this question quietly brought those four "dead companies" back into the spotlight.

III. Beneath the Surface

When Nokia sold its mobile phone business, it took away something that no one noticed: patents.

But after that, its days were not easy. Microsoft acquired Nokia's mobile phone business for $7.2 billion, but only 15 months later, it wrote down $7.6 billion, losing more than the purchase price.

In 2016, Microsoft sold its feature - phone business and manufacturing assets to Foxconn for $350 million, less than 5% of the acquisition price. The brand and smartphone license were taken over by Finland's HMD Global.

Nokia watched all this happen without saying a word. It quietly did one thing. With the cash from selling its mobile phone business, it acquired Alcatel - Lucent for 15.6 billion euros in 2016, integrating decades of communication infrastructure accumulation from Bell Labs into its own territory.

Nokia completely transformed from a mobile phone company into a communication infrastructure company.

It took a full seven years for this transformation to start seeing returns. The 7,000 5G standard - essential patent families became a toll gate for the entire industry. Every 5G mobile phone sold globally has to pay a patent fee to it. No consumer thinks of Nokia when buying a mobile phone, but Nokia takes a share of every transaction.

Nokia retreated from a terminal player on the connection main axis to the basic layer of the connection main axis. It didn't compromise after failure but chose to actively move towards the root of the industry.

In 2025, Nokia's annual revenue was 153 billion yuan, with patent revenue contributing nearly 10%. Its comparable operating profit decreased by 22% year - on - year, but the gross profit margin of its patent licensing business exceeded 80%, making it a typical and most stable cash cow in the industry.

The path Huawei took was completely opposite to Nokia's. It didn't retreat downwards but dug inwards.

After TSMC cut off the supply, Huawei faced more than just the inability to buy chips. The entire semiconductor industry chain, from lithography machines to EDA software, almost every node relied on Western technology. This was not a gap but a wall.

Ren Zhengfei said that Huawei landed the plane while it was shot in the fuselage and the engine was on fire.

They spent three years building the Hongmeng operating system from an open - source framework to full - stack self - research, and gradually completed the self - research and development chain of the Kirin chips. They found a foundry solution through SMIC's process and created a glimmer of hope in the uncharted territory.

On August 29, 2023, the Mate60 Pro was quietly put on the shelves without a press conference or pre - heating.

When the first people who bought the phone opened the chip, they found the model of Kirin 9000s. The reaction on the Internet was a collective shock. It was not because the chip was very powerful; it was still several generations behind TSMC's most advanced process. What was shocking was the very existence of the chip.

Huawei completed the transformation from a consumer terminal on the connection main axis to a multi - line parallel development in computing power, connection, and ecosystem in six years.

The road cut off by the sanctions forced the emergence of three new roads.

In 2024, the Mate70 was launched, and Terminal CEO He Gang announced that every chip in the Mate70 could be domestically produced.

In January 2026, Huawei's mobile phones returned to the top in the Chinese market.

What Intel did in the trough was to persevere. It changed three CEOs within six years. Each had a plan, but none solved the fundamental problem. The most disastrous move was in 2021, when "Son of Intel" Pat Gelsinger returned in a savior's posture, investing $200 billion to build a wafer factory and vowing to compete with TSMC. At that time, NVIDIA's market value was still about $700 billion.

Three years later, NVIDIA's market value exceeded $4 trillion, and Intel's market value evaporated by $150 billion during Gelsinger's tenure. He was forced to retire.

But he left something: the foundation of the 18A process technology.

In March 2025, Chen Liwu took over as the ninth CEO. He did three things: cut redundant operations, focus on the core of the CPU, and then opened the door to accept allies. NVIDIA invested $5 billion, Google signed a multi - year large - scale order, and Elon Musk pulled Intel into the Terafab project.

He saw one thing: the CPU is becoming the scheduling center of the entire AI technology stack. GPUs are responsible for parallel computing, and CPUs are responsible for task orchestration. When an intelligent agent system needs to coordinate the invocation of dozens of tools and manage context switching, the role of the CPU changes from a supporting role to a nervous system.

Intel didn't regain the high - ground of the computing power main axis, but it got back to the control center of the computing power main axis. In the first quarter of 2026, its after - hours stock price rose by nearly 20%. (Extended reading: AI brought down Intel, and AI saved Intel )

What's most peculiar about Oracle is that it did nothing in the trough.

In 2020, founder Larry Ellison moved the company's headquarters from Silicon Valley to Texas. Some people interpreted this as an escape from the pressure in Silicon Valley and a signal that the company was moving towards the margins.

Oracle just stood there, doing databases and enterprise software, and was ridiculed as a relic of the previous era.

It seemed to be waiting for something that it didn't even know if it would come.

After the AI era arrived, that something happened. All large - scale models need to store, manage, and adjust data on an unprecedented scale. The database that Oracle had been working on for decades suddenly became one of the most core infrastructures in the AI era.

In 2025, OpenAI signed a five - year contract worth $300 billion with Oracle. Its annual revenue in 2025 was $57.4 billion, and the annual growth rate of its cloud infrastructure exceeded 52%.

Oracle transformed from a "legacy" in the computing power main axis to a key node in the data layer of the AI era. No one invited it back; it was AI that pushed it to the front stage.

IV. Different Paths, Same Destination

If we line up the resurrection methods of these four companies, we can see four completely different paths.

Nokia moved downwards, retreating from the terminal to the basic layer and using patents to turn itself into the foundation of the entire industry.

Huawei moved inwards, transforming the pain of external supply cut - off into self - research capabilities and extending from the connection main axis to computing power, ecosystem, and full autonomy in the automotive field.

Intel waited passively and waited for the structural demand for CPU scheduling capabilities in the AI inference era.

Oracle stayed in place and waited for the explosive demand for databases in the AI era.

The paths are different, but one thing is the same: These four companies didn't wait to die in their original positions but completed a certain form of ecological niche migration.

More importantly, none of their resurrections was by luck. Each company's return was based on years of technological accumulation and R & D investment.

Nokia's patents were developed during its era as the mobile phone king. Intel's CPU architecture is the result of 70 years of precipitation in the entire semiconductor industry chain. Huawei's Kirin chips are the result of three years of high - intensity investment of hundreds of billions. Oracle's database is based on 40 years of product iteration.

These accumulations never disappeared during those silent years. It's just that the outside world couldn't see them.

This is actually a recurring structure in the history of technology.

In the 19th - century railway boom, the market's attention was focused on railway companies. But in the end, those who reaped long - term value were never just infrastructure operators. Instead, they were enterprises that re - established the business order based on the railway network, such as Standard Oil and AT & T.

After the Internet bubble burst, communication giants like Lucent became a thing of the past. However, the vast optical fiber backbone network laid in excess by the industry unexpectedly remained and became the digital highway shared by the whole people in the cloud - computing and mobile eras.

Each technological wave will give rise to a group of companies that are temporarily overshadowed. Their core capabilities, which are not needed in the current cycle, will not disappear out of thin air. When the industrial paradigm shifts and new demands explode, the shadow over them will be completely lifted.

There are only two types of enterprises that are truly and permanently eliminated: Those with shallow accumulation, only relying on short - term era dividends, or betting on obsolete tracks that are destined to be replaced.

Kodak is one, Blockbuster is one, and Yahoo is also one.

Nokia, Intel, Huawei, and Oracle belong to a completely different category. Their long - term accumulation in the underlying technology, which seems useless in the trough of the old - to - new cycle transition, becomes the indispensable key foundation for the entire industry in the next technological revolution.

V. The Winners and the Losers

In a rapidly changing industrial cycle, all eyes will be on the winners.

But what is more worth asking is: Where did the losers go?

Nokia lost in the mobile phone market and moved to the patent layer and infrastructure layer. Intel lost in the AI training era but waited for the scheduling center in the inference era. Huawei lost its position in the global mobile phone market but gained ecological autonomy and a