「หุ้นแก่» กลายเป็น «หุ้นใหม่»: จาก Dell ถึง Nokia AI จะประเมินค่าโครงสร้างพื้นฐานเก่าอย่างไร?

If you had said a year ago that Dell, Nokia, Cisco, Corning, and Western Digital would once again become hot targets in AI trading, you’d probably be dismissed as out of touch...

After all, for a long time, when the market talked about AI, the immediate reaction was typically Nvidia, memory, optical modules, power, and data centers. These were either close enough to the GPU or directly in the hottest part of the computing power expansion. In contrast, old-school tech companies like Dell, HP, Nokia, Cisco, Corning, and Seagate were more often labeled as "slow growth," "old stories," and "inflexible valuations."

But surprisingly, this batch of once-unsexy old-school tech stocks has performed quite well recently, prompting the market to start re-evaluating them.

The market quickly and adaptively found a suitable explanation: When AI moves from model parameters to real-world data centers, the market will naturally seek out companies with delivery capabilities and infrastructure expertise. This is the reason Dell, HP, Nokia, and others are being seen again.

So, is this a genuine industry revaluation, or is the market just putting a temporary new narrative on old-school tech stocks?

1. AI Market Shift: Why Revalue Old-School Tech Stocks?

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

This is easy to understand. Whoever has the strongest model and can secure the most GPUs gets the most direct market premium. In this phase, investors were most willing to buy into AI imagination, the computing power supply gap, and core beneficiaries like Nvidia.

But the issue is that AI cannot ultimately just stay in press releases and model parameters. Models need to be trained, requiring data centers; inference needs large-scale deployment, requiring servers, networks, storage, and power; and for enterprises to truly use AI, they need a complete IT architecture and delivery capability.

In other words, AI is not a problem a single GPU can solve; it’s a complete, complex systems engineering project. This is the starting point for the re-pricing of old-school tech companies.

Previously, the market might have seen Dell and thought of PCs and traditional servers; HPE and thought of enterprise hardware; Nokia and thought of the old 5G equipment story; Cisco and thought of traditional networking equipment; Corning and thought of glass and fiber optics; Western Digital and Seagate and thought of hard drive cyclical stocks.

These labels aren't wrong, but in the AI infrastructure cycle, their roles have changed. Building AI data centers requires rack-scale servers, liquid cooling, storage, network switches, fiber optic connections, data management, power support, and enterprise-grade delivery capabilities. The larger the AI cluster, the higher the demands for system integration, network transmission, storage capacity, and operational capabilities.

Therefore, the essence of this revaluation is not that the market suddenly became nostalgic, nor is it that old companies are collectively jumping on the AI bandwagon. It's that as AI enters the phase of orders, revenue, and delivery, the market is starting to look for who can *actually build the AI infrastructure*.

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

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

2. From Servers to Networks to Storage: Old-School Tech Stocks Enter the AI Infrastructure Chain

Overall, the old-school tech stocks being revalued by AI can be broadly categorized into three lines: Servers & System Integration, Networking & Connectivity, and Storage & Data Management.

The first line is Servers and System Integration.

Dell is the most typical example here. In its latest quarterly report, Dell delivered very strong data: Q1 FY27 revenue reached $43.8 billion, AI orders hit $24.4 billion, and it confirmed $16.1 billion in AI server revenue. The company also raised its full-year FY27 AI server revenue forecast to $60 billion and its midpoint annual revenue guidance to $167 billion.

This data set is important because it changes how the market views Dell. Previously, investors looked at Dell and focused on PC cycles, traditional servers, and enterprise hardware demand. But now, the market is starting to see if Dell can become the general contractor for building AI factories.

Its advantage isn’t making its own GPUs but its supply chain, delivery capabilities, enterprise customer relationships, server system design, and ability to integrate with the Nvidia ecosystem. An AI server isn't just selling a GPU; it needs to be installed in a rack, connected to networking, power, and liquid cooling systems, and then delivered to cloud providers and enterprise customers.

Dell captures this crucial step from chip to system deployment. HPE's logic is similar.

HPE's stock surged after its latest earnings report, primarily due to strong AI infrastructure demand. The company's Q2 revenue reached $10.68 billion, a 40% year-over-year increase; revenue from cloud & AI-related businesses hit $7.71 billion, and it raised its full-year FY2026 growth forecast. More importantly, HPE has added networking capabilities through its Juniper acquisition, transforming it from a traditional server company into more of an "AI networking + enterprise infrastructure" platform.

So, the revaluation logic for Dell and HPE isn't "they will become the next Nvidia," but rather, they are becoming crucial system integrators within the AI factory construction team.

The second line is Networking and Connectivity.

One of the most overlooked aspects of AI infrastructure is connectivity. Computing power doesn't exist in isolation. Data centers need high-speed internal interconnects, fiber optic connections between data centers, and as AI applications move to the edge and end devices, stronger telecom networks and wireless infrastructure. The larger the scale of AI training and inference, the more networking and connectivity cease to be supporting roles and become critical infrastructure determining computing efficiency.

This is why Corning, Nokia, and Cisco are being discussed again by the market. Corning is a classic example. It's not a traditional AI chip stock, but its fiber optics, optical connectivity, and optical communication materials are essential supporting components for AI data center expansion.

The company's Q1 2026 core sales reached $4.35 billion, an 18% year-over-year increase; its optical communications business sales hit $1.846 billion, a 36% increase year-over-year. The company noted that demand for Gen AI products and new long-term agreements with large hyperscale customers are key growth drivers. This proves AI data centers don't just need GPUs; they also need the fundamental materials to actually connect the computing power.

Nokia's story extends from traditional 5G equipment to AI-RAN, 6G, and AI-native wireless networks. Nvidia previously announced a $1 billion investment in Nokia, collaborating to advance AI-RAN and the transition from 5G to 6G. This signal is significant because AI traffic won't just stay in data centers; it will also reach devices like phones, cars, robots, and AR/VR headsets. As long as AI applications continue to spread to the edge and mobile networks, telecom infrastructure companies will regain narrative space.

Cisco's logic leans more towards data center networking. The company's Q3 FY2026 revenue was $15.8 billion, a 12% year-over-year increase; data center switching orders grew over 40% year-over-year. Remember, in an AI cluster, the network isn't just simple cabling; it's a critical factor affecting data transmission efficiency, GPU utilization, and cluster stability.

The common logic for this category of companies is: the more AI scales towards deployment, the more valuable networking and connectivity become.

The third line is Storage.

This line has become widely known in the market over the past two months. AI doesn't just lack computing power; it also lacks storage. While the market previously focused on HBM, DRAM, and NAND, now high-capacity HDDs are back in the spotlight. AI model training, inference logs, video data, enterprise data, and cold data archiving all generate massive storage capacity needs.

Western Digital is a key representative. The company's latest quarterly revenue grew 45% year-over-year to $3.34 billion and provided next-quarter revenue guidance above market expectations. More importantly, the market noted that demand for its high-capacity hard drives is primarily driven by AI and cloud data centers. Seagate is similar, benefiting significantly from high-capacity nearline hard drives, with an increasing proportion of revenue coming from data center customers.

Of course, not all data in the AI era needs to sit on the most expensive, high-speed storage. Massive amounts of cold data, training data, log data, video data, and archival data still require cost-effective, high-capacity hard drives. So, the revaluation logic for WDC and STX isn't "hard drives are suddenly revived," but rather that the AI data explosion has made storage a non-negotiable necessity again.

3. What Constitutes a Real Revaluation?

However, old-school tech stocks being revalued by AI doesn't mean all old companies are blindly worth bullish bets.

A crucial distinction exists here: some companies are genuinely integrated into the AI infrastructure chain. To judge if a company is truly being revalued, one should look at at least three criteria:

• First, are there orders and revenue realization? For example, Dell's AI orders and AI server revenue, HPE's cloud & AI business, Corning's optical communications revenue, Cisco's data center switch orders, and WDC's high-capacity hard drive demand – these matter more than just telling an AI story.
• Second, have guidance estimates been raised? If AI stays only in press releases and product introductions, stock prices can easily rise and then fall. But if management is willing to raise full-year revenue expectations, business growth forecasts, or key product shipment estimates, it signals that AI demand isn't just short-term sentiment; it might be changing the company's growth trajectory. This is why the market is re-pricing companies like Dell and HPE.
• Third, can profit quality keep up? The biggest problem for old-school hardware companies has always been gross margins and cyclicality. Fast growth in AI server revenue doesn't necessarily mean high profit elasticity. A storage price increase could be just a short-term supply-demand mismatch. Increased networking equipment orders need to translate into sustainable profits.

A truly good revaluation should involve a simultaneous improvement in revenue growth, order visibility, and profit quality.

If revenue is up but gross margins are squeezed thin, or if demand is just a short-cycle restocking event, then the valuation revaluation will be limited. Ultimately, the market isn't buying "old companies telling new stories"; it's buying "old assets combined with new demand being converted into new profits."

This is the most critical point to watch in this round of "old trees blooming new flowers." AI won't turn all traditional tech companies back into growth stocks. It will only select those truly positioned at critical infrastructure bottlenecks and capable of converting AI demand into orders, revenue, and profits.

Final Thoughts

Objectively speaking, the AI market cycle has moved beyond just "who has the stronger model" or "who has more GPUs." The real change is that AI is entering a genuine construction phase.

As AI data centers continue to multiply, server companies will be re-priced. As computing clusters become more complex, networking companies will be re-priced. As data centers demand more fiber optic connections, materials companies will be re-priced. As AI data continues to explode, storage companies will be re-priced.

This is why old-school tech stocks are being re-examined by the market. They haven't suddenly become young again; rather, the AI era has a renewed need for the infrastructure they provide.

But this also means the revaluation won't be evenly distributed among all "old-economy tech stocks."

Only those old-school tech companies that can genuinely tap into the capital expenditure chain of data centers and enterprise deployments have the potential to move from "valuation recovery" to "logical revaluation."