After AI enters the infrastructure construction phase, these companies that were once considered low-growth—servers, networks, fiber optics, and storage—are seeing "old trees blossom anew."

Written by: Jim, MSX Maitong

One year ago, if someone told you that Dell, Nokia, Cisco, Corning, and Western Digital would become hot stocks in AI trading again, you would probably think they were a bit unclear...

After all, for a long time, when the market talked about AI, the first reaction was usually Nvidia, storage, optical modules, power, and data centers. They were either closely related to GPUs or directly involved in the hottest segments of computing power expansion. In contrast, traditional tech companies like Dell, HP, Nokia, Cisco, Corning, and Seagate were more often labeled as "slow growth," "old stories," and "inelastic valuations."

Yet, these seemingly unsexy traditional tech stocks have recently shown impressive performance, prompting the market to start discussing them again.

The market quickly found an appropriate explanation: When AI shifts from model parameters to real data centers, the market will naturally seek out companies that have delivery capabilities and infrastructure capacities. This is the reason why Dell, HP, Nokia, and others are being recognized again.

So, is this a genuine industrial revaluation, or is it just a new narrative temporarily wrapped around traditional tech stocks by the market?

1. The AI Trend Shift: Why Revalue Traditional Tech Stocks?

In the past few years, the core clue of AI trading has been very clear: first look at the models, then look at the computing power.

This is easy to understand; whoever has the strongest model and can acquire the most GPUs will receive the most immediate market premium. During this phase, what investors are most willing to buy is the imagination of AI, the gap in computing power supply, and core beneficiaries like Nvidia.

However, the problem is that AI cannot be limited to press conferences and model parameters. After all, models need to be trained, requiring data centers; inference needs to be implemented on a large scale, needing servers, networks, storage, and power; for companies to truly use AI, a complete IT infrastructure and delivery capability are necessary.

In other words, AI is not a problem that can be solved with just a GPU; it is a complex system engineering task, which is also the starting point for the revaluation of traditional tech companies.

In the past, when examining Dell, one might think of PCs and traditional servers; looking at HPE would evoke thoughts of enterprise hardware; thinking of Nokia would bring to mind outdated stories of 5G devices; Cisco would remind one of traditional network equipment; Corning is thought of in terms of glass and fiber optic materials; and Western Digital and Seagate would bring up hard disk cyclical stocks.

These labels are not incorrect, but their roles have changed in the AI infrastructure cycle—AI data centers need to be built, requiring rack servers, liquid cooling, storage, network switches, fiber optic connections, data management, power support, and enterprise-level delivery capabilities. The larger the AI clusters, the higher the requirements for system integration, network transmission, storage capacity, and operational maintenance capabilities.

Therefore, the essence of this revaluation is not that the market suddenly became nostalgic, nor that old companies are collectively riding the AI wave; it is that as AI enters the ordering, revenue, and delivery phases, the market starts to search for "who can truly build AI infrastructure."

These companies may not be the most attractive, but they share a common advantage: the customer relationships, channels, supply chains, delivery experience, and infrastructure capabilities accumulated over the past several decades are becoming valuable again during the large-scale deployment phase of AI.

In other words, AI is re-evaluating a group of "old assets" within the context of "new demand."

2. From Servers, Networks to Storage: Traditional Tech Stocks are Being Integrated into the AI Infrastructure Chain

Overall, this round of AI revaluation of traditional tech stocks can be roughly divided into three lines: servers and system integration, networking and connectivity, and storage and data management.

The first line is servers and system integration.

Dell serves as the most typical example. In its latest quarterly report, Dell presented very strong data: Q1 FY27 revenue reached $43.8 billion, AI orders totaled $24.4 billion, and confirmed $16.1 billion in AI server revenue. The company also raised its full-year AI server revenue forecast for FY27 to $60 billion, increasing the median revenue guidance for the full year to $167 billion.

This set of data is significant because it changed how the market views Dell. Previously, investors looked at Dell mainly through the lens of the PC cycle, traditional server demand, and enterprise hardware. But now, the market is looking at Dell to see whether it can become a general contractor in AI factory construction.

Its advantage is not in making GPUs, but in its supply chain, delivery capability, enterprise customers, server system design, and compatibility with the Nvidia ecosystem. An AI server isn't just sold with a single GPU; it needs to be installed into a cabinet, connected to the network, power, and liquid cooling systems, and delivered to cloud vendors and enterprise customers.

Dell is benefiting from this critical phase from chip to system deployment, and HPE’s logic is similar.

HPE's stock surged after its latest earnings report, driven by strong demand for AI infrastructure. The company’s Q2 revenue hit $10.68 billion, up 40% year-over-year; and its revenue from cloud and AI-related businesses reached $7.71 billion, with upward revisions to growth expectations for FY2026. More importantly, HPE has also added network capabilities through Juniper, transforming it from just a traditional server company into more of an "AI network + enterprise infrastructure" platform.

Hence, the revaluation logic for Dell and HPE does not imply "they want to become Nvidia"; rather, they are becoming very important system integrators in the AI factory construction teams.

The second line is networking and connectivity.

One of the easily overlooked aspects of AI infrastructure is connectivity. Computing power does not exist in isolation. Data centers need high-speed interconnections internally; fiber optic connections are needed between data centers, and as AI applications move towards the edge and endpoints, stronger telecommunications networks and wireless infrastructures are necessary. The larger the scale of AI training and inference, the more network and connectivity become critical infrastructure that dictates computing efficiency.

This is also the reason why Corning, Nokia, and Cisco are back in market discussions. Corning is a typical example; it is not an AI chip stock in the traditional sense, but its fiber optics, optical connections, and optical communication materials are vital for the expansion of AI data centers.

The company's Q1 2026 core sales reached $4.35 billion, a year-on-year increase of 18%; among which, optical communication business sales reached $1.846 billion, a 36% increase year-on-year. The company also mentioned that demand for Gen AI products and new agreements with large hyperscale clients are significant growth drivers, indicating that AI data centers not only need GPUs but also the basic materials that genuinely connect computing power.

Nokia's narrative extends from traditional 5G devices into AI-RAN, 6G, and AI-native wireless networks. Nvidia previously announced a $1 billion investment in Nokia, with both parties collaborating to promote AI-RAN and transition from 5G to 6G. This signal is crucial because AI traffic in the future will not only stay in data centers; it will also enter mobile devices, cars, robots, and AR/VR terminals. As long as AI applications continue to spread towards the edge and mobile networks, telecommunications infrastructure companies will regain narrative space.

Cisco's logic leans more towards data center networking; its Q3 FY2026 revenue reached $15.8 billion, a 12% year-on-year increase; data center switch orders grew over 40% year-on-year. In AI clusters, networks are not merely connecting lines; they are critical elements that affect data transmission efficiency, computing power utilization, and cluster stability.

The common logic among these types of companies is that as AI moves towards large-scale deployment, networks and connectivity become more valuable.

The third line is storage.

This sector has been widely acknowledged in the market over the past two months, meaning that AI not only lacks computing power but also storage. Previously, the market focused on HBM, DRAM, and NAND, but now high-capacity HDDs have re-entered the spotlight because AI model training, inference logs, video data, enterprise data, and cold data archiving will all lead to greater storage capacity demands.

Western Digital is one representative in this line. The company’s latest quarterly revenue grew by 45% year-on-year to $3.34 billion, providing guidance for the next quarter higher than market expectations. More importantly, the market has noted that the high-capacity hard disk demand mainly comes from AI and cloud data centers; Seagate is similar, significantly benefiting from high-capacity nearline hard drives, with an increasing percentage of data center clients.

Of course, the AI era does not mean all data should be stored in the most expensive high-speed storage. A significant amount of cold data, training data, log data, video data, and archived data still requires cost-effective high-capacity hard drives, so the revaluation logic for WDC and STX is not that "hard disks suddenly revived," but rather that the explosion of AI data has made storage a necessity again.

3. What Constitutes a True Revaluation?

However, the revaluation of traditional tech stocks by AI does not mean that all old companies are worthy of a brainless bullish outlook.

The most significant distinction is that some companies have genuinely entered the AI infrastructure chain, so to determine whether such companies have been truly revalued, at least three criteria must be considered:

• First, are there any orders and revenue realization: For example, Dell's AI orders and AI server revenue, HPE's cloud and AI-related business, Corning's optical communication revenue, Cisco's data center switch orders, WDC's high-capacity hard disk demand—all of these are more critical than simply telling AI stories;
• Second, is there an upward revision in guidance: If AI remains confined to press conferences and product launches, it is easy for stock prices to spike and then retreat. However, if management is willing to raise full-year revenue expectations, business growth forecasts, or key product shipment expectations, it indicates that AI demand is no longer just a short-term sentiment but may be changing the company’s growth trajectory, which is why the market is re-pricing companies like Dell and HPE;
• Third, can profit quality keep pace: The biggest issue for traditional hardware companies has always been gross margin and cyclicality. Rapid growth in AI server revenue does not necessarily equate to high-profit elasticity; rising storage prices may just be a short-term mismatch of supply and demand; increased orders for networking equipment must also be assessed for their ability to convert into sustained profits;

True revaluation should see improvements in revenue growth, order visibility, and profit quality together.

If only revenue rises but gross margins are squeezed thin, or if demand is simply a short cycle of inventory replenishment, then the valuation revaluation will be limited. Ultimately, the market buys not "old companies telling new stories," but rather "can old assets plus new demands translate into new profits."

This is also what makes this round of "old trees blossoming anew" particularly noteworthy: AI will not turn all traditional tech companies back into growth stocks; it will only filter out those that are truly stuck at critical infrastructure junctures and can transform AI demand into orders, revenues, and profits.

In Conclusion

Objectively speaking, the AI trend has reached a point where it is no longer just about "who has the stronger model" or "who has more GPUs"; the real change lies in AI entering the phase of actual construction.

As more AI data centers are built, server companies will be re-priced; as computing power clusters become increasingly complex, networking companies will be re-priced; as data centers require more fiber optic connections, materials companies will be re-priced; as AI data continues to explode, storage companies will also be re-priced.

This is why traditional tech stocks are being re-seen by the market: they are not suddenly rejuvenating; rather, the AI era requires the infrastructures they hold once again.

But this also means that this round of revaluation will not be evenly distributed among all "old tech stocks."

Only those that can truly enter the capital expenditure chain for data center and enterprise deployment will have the potential to transition from "valuation repair" to "logical revaluation."