Written by: Xiaobing, Deep Tide TechFlow

In 2014, Microsoft spent 7.2 billion dollars to buy Nokia's mobile business.

That year, everyone thought the story of this Finnish company, which had been around since 1865, was over. The Nokia 3310, which could crack walnuts, was tough enough to be thrown against the wall, and was famous for playing Snake, along with the entire company behind it, was nailed into the album of "tears of the times."

Eleven years later, Jensen Huang called and said he wanted to invest 1 billion dollars in them.

Nokia's stock has risen by about 73% from the beginning of January this year to now, an increase of 130% compared to the same period last year.

This is not a "dead cat bounce" of an "old tech stock," but an undervalued hidden gem in the entire AI narrative from 2025 to 2026.

Moreover, almost no Chinese investors are seriously discussing it.

Who is Justin Hotard?

The story begins with a name.

In February 2025, Nokia's board announced that the current CEO Pekka Lundmark would step down, and on April 1, a man named Justin Hotard from the United States would take over.

This is the first American-born CEO of Nokia since its founding in 1865.

Hotard is a typical "not very famous but always stepping on the right opportunities" person in the tech circle. He has a bachelor's degree in electrical engineering from the University of Illinois and an MBA from MIT Sloan. He worked at Hewlett Packard Enterprise (HPE) for eight and a half years, finally becoming the head of high-performance computing and AI laboratories, delivering the world's first exascale supercomputer to the U.S. Department of Energy. Then, in early 2024, he was poached by Intel to oversee the data center and AI division, reporting directly to Pat Gelsinger.

Note his resume. HPC, data centers, AI—these three key terms have nothing to do with "Nokia" and its image over the past decade.

What does Nokia do? They manufacture base stations, telecommunications equipment, fiber optics, and sell to operators. A typical, slow, "old Europe" hardware company that has been forgotten by the market.

But in early 2025, Nokia's board made a seemingly irrational decision: they did not want someone who understood telecommunications; they wanted someone who understood AI.

In the appointment announcement, the board chair Sari Baldauf stated: "The AI and data center markets are key areas for Nokia's future growth."

At that moment, almost no one took it seriously. The market reacted flatly, with a slight rise in stock prices. All analysts were writing "Can the newcomer turn around the Finnish old factory?" types of balanced analyses.

No one realized that this company was quietly changing its engine.

An Undervalued Acquisition

If you only look at Hotard's appointment, it's just a normal executive rotation. But when you put it together with another event from six months prior, the narrative completely changes.

In June 2024, Nokia announced a 2.3 billion dollar acquisition of an American company—Infinera.

What is Infinera? It specializes in optical networking, which, simply put, refers to "fiber optic communication equipment" between data centers and between racks.

If you have talked to people in AI infrastructure, you know a fact:

The biggest bottleneck in AI data centers is not GPU, it's optical communication.

NVIDIA can cram 72 GPUs into one rack, and these GPUs need to exchange data frantically. A data center can have tens of thousands of GPUs, and they also need to exchange data. Two data centers must sync training data. For every additional cluster, the demand for optical modules grows exponentially.

That's why in the past two years, optical module companies like Coherent in the U.S., Zhongji Xuchuang in China, and Newpower have seen their stock prices fly.

And Infinera is one of the few companies that master both photonic integrated circuits (PIC) and internal data center interconnect technologies. It already has established customer relationships with large-scale cloud vendors in North America such as Microsoft, Amazon, and Google.

When Nokia signed this deal in June 2024, the market interpreted it as: "A traditional telecom company has acquired another traditional company in fiber optics," a typical "two elephants cuddling together to keep warm" narrative.

But when the transaction closed in February 2025, the financial numbers of this old Finnish company began to change:

• In the full year of 2025, optical networking revenue grew by 17% year-on-year.
• In the first quarter of 2026, optical networking sales reached €821 million, a 20% year-on-year increase, surpassing IP and core software to become Nokia's second-largest business.
• Sales contributed by AI and cloud clients surged 49% year-on-year in a single quarter.

The most crucial number: In the first quarter of 2026, AI and cloud clients placed orders worth 1 billion euros with Nokia.

What does this number mean? It corresponds to a quarterly order volume exceeding Infinera's total sales for the entire year before the acquisition.

And all of this went almost unnoticed outside of Western tech media.

Jensen Huang's Call

What truly sent the market into a frenzy was October 28, 2025.

On that day, NVIDIA announced at the GTC conference in Washington: NVIDIA will invest 1 billion dollars in Nokia at a price of $6.01 per share.

Pay attention to this detail, "at $6.01 per share," this is a subscription price, not the market price. NVIDIA is not buying stocks on the secondary market; Nokia specifically issued a new batch of shares for it, and NVIDIA's investment is strategic, not just financial.

Why does Jensen Huang want to give Nokia 1 billion dollars?

NVIDIA's official statement is: The two sides will co-develop AI-RAN (AI Radio Access Network). Nokia's 5G and 6G software will be ported to NVIDIA's CUDA platform; NVIDIA's Arc-Pro accelerators, specially developed for the telecom industry, will be embedded in Nokia's base stations.

T-Mobile in the U.S. will be the first pilot operator. Dell will provide servers.

It sounds like another common "AI empowers XX industry" story. But the real twist lies in a technical detail that 99% of people won't notice.

First, you need to know a background: In the AI-RAN space, Nokia is not the only player. Its biggest competitor is Ericsson, also from the Nordics.

What Ericsson and Nokia do appears identical: providing operators with 5G/6G base station equipment. But when it comes to "how to integrate GPU into base stations," they have taken two completely opposite paths.

These two paths are half-jokingly referred to by engineers as "religious wars."

The first path is called Lookaside (bypass acceleration). This is the route taken by Ericsson and Intel. In simple terms: the CPU in the base station remains the primary controller, while the GPU is merely an "assistant" placed beside it. When a computational task needs acceleration, the CPU "sends" the task to the GPU, which returns the results once completed. Data needs to bounce back and forth between the CPU and GPU.

The second path is called Inline (in-line acceleration). This is the route taken by Nokia and NVIDIA. In simple terms: the network data received by the base station first goes to the GPU, is processed there, and then passed to the CPU. The GPU becomes the protagonist while the CPU takes a backseat.

Does it sound like just an engineering sequence issue?

No, this is a fundamental disagreement about "who will be the center of computing in the future."

The entire existence of NVIDIA hinges on proving that the GPU should be the center of data processing, pushing the CPU to the side. The design philosophy of the CUDA ecosystem is "GPU-centric." The Lookaside approach fundamentally presupposes that "the CPU is still the boss," which is incompatible with NVIDIA's worldview.

Thus, when NVIDIA sought a telecom partner, it could not choose Ericsson. It had to pick a partner willing to place the GPU in the central position.

Nokia is that partner.

That's why this 1 billion dollars is not just a simple "strategic investment." Jensen Huang has placed a stamp on a new map of AI narrative, he is buying an entry point that allows NVIDIA's GPUs to penetrate into 5 million base stations worldwide.

According to analysis firm Omdia, it is predicted that by 2030, the cumulative market size of AI-RAN will exceed 200 billion dollars.

If told correctly, this 1 billion dollars from Jensen Huang could be one of the highest return investments of his life.

Geopolitical Factors Helped

Nokia's comeback has sensitive hidden lines.

On April 13, 2026, Oliver Wong, an analyst at Bank of America, upgraded Nokia from "neutral" to "buy," raising the target price from €6.87 significantly to €10.70. On that day, Nokia's stock surged by 9.67%, with trading volume enlarging by 178% compared to the three-month average.

In that report, Oliver Wong listed four reasons for Nokia's undervaluation. Among them, the third reason was phrased very delicately but was very clear:

"After European countries gradually restrict Huawei and ZTE, Nokia has effectively become 'the last available Western sovereign-grade supplier.'

In simple terms: for Europe to build sovereign data centers and sovereign 5G/6G networks, Chinese equipment cannot be used, and there are no such companies based in the U.S., meaning the only options left for Western suppliers are Nokia and Ericsson. But Ericsson lacks full-stack capability in optical networking, and with Infinera acquired by Nokia, while Cisco is also an American company, the funding for Europe's sovereign cloud will almost invariably flow to Nokia.

This is a classic "geopolitical arbitrage" opportunity, a change in the international order has given Nokia a huge gift; as long as they remain on the track, they can reap this dividend.

Furthermore, the demand for optical networking from large American cloud companies, and T-Mobile's bet on AI-RAN, three streams of funding are simultaneously directed toward Nokia from three different directions.

The Market Took 18 Months to React

Putting all the clues together, you will see a very dramatic timeline:

• In June 2024, Nokia announced the acquisition of Infinera.
• In February 2025, Hotard was appointed as the new CEO.
• In October 2025, NVIDIA invested 1 billion dollars.
• On April 13, 2026, Bank of America upgraded Nokia, and stock prices rose 9.67% in one day.
• On April 22, 2026, the Q1 financial report disclosed 1 billion euros in AI/cloud orders, optical networking business +20%.
• On April 27, 2026, CFRA directly doubled the target price from the original $8 to $16, and Nokia's stock price hit a new high since 2015.

Did you notice?

The fundamentals began to change 18 months ago. But the market took 18 months to connect these dots.

This is the classic "value discovery" process. When a story hasn’t been clearly told, everyone treats it as "old wine in new bottles"; by the time the story is clear, much of the valuation has already been repaired.

Nokia's current price-to-earnings ratio is 26 times the forward PE, which is not expensive for a 17% growing optical networking business. However, compared to the low points earlier this year, it is no longer that "forgotten stock" lying on the ground.

In the past two years, Chinese investors have been focused on NVIDIA, TSMC, Broadcom, and AMD, the engines of this AI wave.

But beyond the engine, there are gearboxes, drive shafts, tires, and highways.

The AI narrative is spreading from "chips" to "pipelines."

The story of optical module manufacturers has been told for over a year; the next to be revalued by the market may be the base stations, optical fibers, power for data centers, and cooling systems.

The story won’t repeat, but the story will rhyme.

When a new technological paradigm truly arrives, the biggest Alpha might not be in the obvious places.

It is in those corners you think have "already been forgotten."