Why are these "old-timers" in the PC industry outperforming NVIDIA?
Since 2026, a group of "old-timers" from the PC era are making a collective comeback.
Intel has seen a year-to-date increase of 234.74%; Corning has risen by 124.46%; Nokia has climbed 156.22%. Even companies like Dell and Cisco, once regarded as "legacy of the old era" by the market, have suddenly been brought back to the center stage of the capital market.
Their increases not only far outperformed the Nasdaq Index but also significantly exceeded those of core AI assets like NVIDIA and Google.
So, why have these old-timers, long labeled as "outdated," become important again?
Today, let's talk about these "old-timers" who have been brought back to the table.
Intel: Back at the AI Table
A year ago, Intel was the typical example of a company that "missed the AI wave" in the eyes of Wall Street.
In April 2025, its stock price dropped to $17.665, hitting a ten-year low, and its market value shrank to less than $70 billion.
At that time, the market had almost given Intel a "death sentence with a reprieve."
Everyone was talking about GPUs, NVIDIA, and CUDA. Intel, on the other hand, was like the old king abandoned by the era.
But no one expected that just one year later, Intel's stock price soared above $120, a rise of over 579%, and its market value reached $600 billion again.
The core of its turnaround lies in a statement that many were reluctant to admit in the past: The further AI develops, the more important the CPU becomes.
In the era of large models in the past, it was essentially the "training era."
At that time, the GPU was the absolute protagonist. The CPU in the server was more like a supporting role, and the CPU/GPU ratio once reached 1:8.
But in 2026, AI started to move from "chatting" to real work.
Especially after the explosion of AI Agents, models are no longer just answering questions but starting to break down tasks, call tools, search for information, run code, and schedule systems like humans.
Many of these tasks are not what GPUs are good at.
It is the CPU that is truly responsible for system scheduling, toolchain management, and task orchestration.
A study by the Georgia Institute of Technology shows that in the Agent workflow, CPU-side tool processing may account for 50% - 90% of the total latency.
Intel CEO Chen Liwu even directly said at the earnings conference: In the past, it was one CPU paired with 7 - 8 GPUs, but now it has become 1:4. In the future, after AI Agents are truly popularized, it may even reach 1:1, or even the CPU may dominate.
This has brought great opportunities to Intel. You know, Intel has a market share of over 70% in the global CPU market.
Even more surprisingly, since the beginning of this year, server CPUs have even been in short supply. In January this year, there were reports that Intel's server CPU production capacity for the whole year of 2026 was basically sold out.
Since March, the prices of server CPUs have generally increased by 10% - 20%. High-end AI server products with 64 cores and 96 cores are especially in short supply, and the price increase in some channels has even exceeded 20%.
The market expects this shortage to last until 2027.
Meanwhile, another business of Intel that has been ridiculed for many years - wafer foundry - is also starting to turn around.
As TSMC's 3nm and 5nm production lines are fully loaded for a long time, and the anxiety about local manufacturing in the United States is growing stronger, the market is increasingly convinced that the United States really needs a "local version of TSMC."
And the only company capable of taking this position is almost Intel.
So, giants like Apple, Google, and NVIDIA have begun to gradually shift some orders to Intel.
On May 9th, after Apple announced that it would give some chip foundry orders to Intel, the company's stock price skyrocketed by nearly 14% on the same day.
So, the market began to realize that Intel may be the most undervalued part of AI infrastructure.
Just like this, after falling behind for two years, this giant, once considered to have "missed the AI era," has finally returned to the core table of AI.
The "Glass" Company Bet on by NVIDIA: Corning's Stock Price Doubled in Less Than Six Months
Corning, which once produced glass bulbs for Edison, has suddenly become one of the hottest companies in the AI circle recently.
Since the beginning of the year, Corning's stock price has soared from $87.39 to $196.17, a rise of over 120%.
Corning's sharp rise is due to its optical fibers made of special glass. This material directly addresses the physical bottleneck of AI computing power expansion: data transmission.
To put it simply, GPUs are like neurons in the AI brain.
And the optical fibers produced by Corning are the "nerve fibers" connecting these neurons. The more GPUs there are and the larger the AI cluster is, the more terrifying the data transmission pressure becomes.
According to Corning's data, an AI server equipped with 32 computing chips requires 4 times as much optical fiber as a traditional cloud server; for an AI cluster unit with 72 GPUs, the optical fiber demand is even 16 times that of a traditional rack.
That is to say, the more GPUs are piled up in AI, the more the world depends on optical fibers.
So, Corning has suddenly transformed from an "old glass company" into a core player in AI infrastructure.
On May 6th, 2026, NVIDIA announced that it would invest up to $3.2 billion in Corning to build three new optical manufacturing plants in the United States, specifically for AI "optical connections."
On the day the news was announced, Corning's stock price soared by over 20% during intraday trading and finally closed up 12%.
Actually, Meta had already taken action before NVIDIA.
Meta signed a long-term agreement worth $6 billion with Corning to supply fiber optic cables for its AI data centers.
The crazy purchasing by these two tech giants has also made Corning's optical communication business take off completely.
According to the earnings report released at the end of April, Corning has achieved year-on-year sales growth for eight consecutive quarters.
In the first quarter of fiscal year 2026, the company's revenue increased by 20% year-on-year to $4.144 billion, and its net profit skyrocketed by 136% year-on-year to $371 million. The core growth engine is optical communication.
This is the most amazing thing about this company.
More than 100 years ago, it made light bulb glass for Edison; more than 100 years later, it has become the "nerve fiber supplier" in the AI world.
Sometimes you'll find that although the times change, some companies can always find their place in the next technological revolution.
Soaring Over 120%, Nokia Becomes a Newcomer in the AI Network
Nokia's sharp rise this time is actually a very magical thing.
From $6.47 at the beginning of the year to $16.46 now, Nokia's stock price has more than doubled in half a year, with a rise of 156.22%, even hitting a new high in the past 16 years.
The problem is that its fundamentals are not actually very impressive.
In 2025, Nokia's annual revenue only increased by 3%, and its net profit even declined by nearly 50%.
If it were in the past, the market probably wouldn't even take a second look at such an earnings report.
But this is the most absurd thing about the AI era. Sometimes, what determines the stock price is no longer "how much money you're making now" but "whether you're on the main track of AI infrastructure."
If you break down Nokia's business, you'll find that the core of its sharp rise is not its traditional communication business.
In 2025, Nokia's revenue from its traditional mobile network business was 7.706 billion euros, a year-on-year decline of 4%, and its operating profit margin was only 2.8%.
It has almost been regarded as a "sunset business" by the market.
But on the other hand, its new business has started to take off completely.
In 2025, Nokia completed the acquisition of the US optical network equipment manufacturer Infinera. After the merger, the revenue from its optical network business directly soared to 3.019 billion euros, a year-on-year increase of 85%.
And this is the real reason why the market has started to reprice Nokia.
If Corning sells the "nerve fibers" in the AI world, then Nokia sells the entire "neural network" composed of these fibers.
As AI has developed to this point, an increasingly real problem has emerged:
There are so many GPUs that a single data center can't hold them all.
Today's large model training has entered the "ten-thousand-card" era. But the power supply and cooling capabilities of a single data center are approaching their physical limits.
A super cluster composed of 100,000 H100s consumes about 1.59 terawatt-hours of electricity per year.
How absurd is this number? It's almost equivalent to the annual electricity consumption of a medium-sized prefecture-level city.
So, AI computing power has to start "distributed survival."
Training is no longer just a matter within one computer room but a large-scale collaboration between multiple regions and multiple data centers. And the most core thing behind this is the high-speed optical network.
Coincidentally, this is exactly the area where Nokia excels.
In the fourth quarter of 2025, Nokia's orders for long-distance transmission in optical networks and IP networks increased by 17%, and the demand from AI and cloud customers increased rapidly.
Meanwhile, there is an even more direct catalyst.
In October 2025, NVIDIA announced a strategic investment of $1 billion in Nokia, holding about 2.9% of the shares. The two sides will jointly promote AI-RAN and the future 6G network architecture.
The once-famous mobile phone giant known for "breaking walnuts" is now starting to weave the "neural network" of the AI world.
Soaring Over 140% in a Year, Dell Becomes the "Arms Dealer" of AI Servers
In this wave of the resurgence of old-tech companies, Dell's comeback is also very dramatic.
Before even half of 2026 has passed, Dell's stock price has risen from $124.95 to $305.08, a rise of 144.17%.
Previously, many people's impression of Dell may have been "that computer-selling company." But in fact, Dell is now one of the largest AI system integrators in the world.
You know, GPUs and CPUs can't work directly on their own. They need to be assembled into complete servers, and they also need to be paired with memory, storage, power supplies, liquid cooling, network systems, and complex software scheduling.
What Dell does is to integrate all the components into a ready-to-use AI server.
In fiscal year 2026, Dell's revenue reached $113.5 billion, a year-on-year increase of 19%, setting a new historical high.
The most remarkable one is the Infrastructure Solutions Group (ISG). Its single-quarter revenue reached $19.6 billion, a year-on-year increase of 73%; the revenue from AI-optimized servers even soared to $9 billion, a year-on-year increase of 342%.
What's even more shocking is the order data. In fiscal year 2026, Dell's total AI orders reached $64.1 billion, and the backlog of orders at the end of the period was as high as $43 billion.
That is to say, it's even a bit "overwhelmed" with work.
For fiscal year 2027, Dell's revenue guidance for AI-optimized servers is $50 billion, a year-on-year increase of 103%.
But Wall Street obviously thinks it's too conservative. Bank of America directly gave a forecast of $60 billion.
Today, about one out of every eight dollars in AI server revenue globally flows to Dell.
Just like this, Dell, once considered a "legacy of the PC era," has instead become one of the most important infrastructure players in the AI era.
Cisco: Soaring 13% Overnight, the "AI Rebirth" of an Old-School Switch
Many young investors may not have seriously looked at Cisco for a long time. After all, for many years, it has been more like a company from the "old era of the Internet."
But since the beginning of the year, Cisco's stock price has risen from $76.21 to $118.33, a rise of over 55%.
What really ignited the market sentiment was the earnings report on May 14th.
On the day the earnings report was announced, Cisco's stock price soared by 17% during intraday trading and finally closed up 13.41%, setting the largest single-day increase since 2011. The company's market value also reached $450 billion again.
The reason why Cisco has become popular is that people have found that as AI develops further, having just GPUs is not enough. The real problem is how GPUs communicate with each other.
Cisco's role in the AI industry chain is actually very much like a "traffic control center in the computing power world."
In an AI super cluster, thousands of GPUs need to process massive amounts of data simultaneously and exchange information in real-time.
And Cisco's switches are essentially responsible for receiving data, distributing data, and then accurately sending the data to the correct location. You can think of it as the "highway scheduling system" in the AI world.
The problem is that AI clusters are becoming more and more exaggerated.
From thousands of cards to tens of thousands of cards, and then to collaborative training across data centers, the entire data traffic is exploding exponentially.
So, the network has started to become one of the most important bottlenecks in AI infrastructure again.
This is also why Cisco's earnings report was so explosive.
The company disclosed that since the beginning of fiscal year 2026, it has won $5.3 billion in AI infrastructure orders from hyperscale customers. Subsequently, it directly raised its annual AI order target from $5 billion by 80% to $9 billion.
Meanwhile, the company expects its AI-related revenue in fiscal year 2026 to reach $4 billion, higher than the previously expected $3 billion.
More intuitively, in the first quarter of 2026, Cisco's orders for data center switches skyrocketed by over 40% year-on-year.
And in order to fully bet on AI, Cisco has even started to "cut off its tail to survive."
The company announced that it would lay off nearly 4,000 employees and