The AI investment cycle is creating opportunities well beyond GPUs. Memory chips, enterprise storage and semiconductor testing equipment are increasingly important as data centres expand to support larger AI models and growing computing workloads.
SK Hynix rose 1.92% after reporting record third-quarter revenue of ₩104.6 trillion and operating profit of ₩83.3 trillion, alongside expanding shipments of next-generation HBM4 memory. Meanwhile, Seagate exceeded expectations with $4.23 billion in revenue, non-GAAP EPS of $7.68 and $1.5 billion in free cash flow. Teradyne also delivered results above expectations.
Together, these developments highlight a broader trend: AI infrastructure spending is creating demand across several parts of the semiconductor and storage supply chain.
But after such strong results, investors face a more difficult question. Are these companies entering a period of sustained earnings growth, or is the market already pricing in the best part of the cycle?
1. SK Hynix: High-bandwidth memory remains the key attraction
SK Hynix has established a strong position in high-bandwidth memory (HBM), a specialised type of memory used alongside advanced processors to handle the enormous data requirements of AI workloads.
As AI models become more demanding, processors need rapid access to large amounts of data. HBM helps address this bottleneck by delivering high bandwidth in a compact package.
This makes memory an increasingly important part of AI infrastructure rather than simply a commodity component.
The expansion of HBM4 shipments is particularly significant because next-generation memory can support evolving accelerator architectures and future AI computing requirements.
However, the reported operating profit of ₩83.3 trillion against revenue of ₩104.6 trillion implies an extraordinary operating margin of approximately 79.6%. If sustained, that would represent exceptional profitability.
Investors should be cautious about assuming these margins will continue indefinitely. Semiconductor memory remains a cyclical industry, and profitability can change as supply expands, competitors introduce new products and customers negotiate pricing.
The key question is whether SK Hynix can maintain its technological advantage while converting strong HBM demand into durable earnings and cash flow.
2. Seagate: AI needs storage as well as computing power
AI discussions often focus on GPUs and memory chips, but data centres also need enormous storage capacity.
Training AI models requires large datasets, while inference and enterprise AI applications generate additional information that must be stored, accessed and managed.
Seagate benefits from demand for high-capacity hard disk drives used in cloud and enterprise storage environments.
Its latest reported results, including $1.5 billion in free cash flow, provide an important reminder that AI infrastructure spending can benefit businesses outside the most prominent chipmakers.
Storage also has a different investment profile from HBM.
High-bandwidth memory is closely linked to advanced AI accelerators, while hard disk drives serve broader data-storage requirements. This creates different demand drivers, competitive dynamics and margin characteristics.
For investors, Seagate’s challenge is to demonstrate that strong pricing, product demand and cash generation can continue as storage capacity expands.
A strong quarter is encouraging, but future returns will depend on whether earnings remain healthy through different stages of the storage cycle.
3. NAND flash: SanDisk is the next important test
SanDisk’s upcoming earnings report could provide another indication of whether the memory recovery is broadening beyond HBM.
NAND flash is widely used in solid-state drives (SSDs), enterprise storage and consumer electronics. Enterprise SSD demand is particularly relevant as businesses upgrade infrastructure to support data-intensive applications.
Unlike HBM, NAND operates in a market where supply growth and pricing cycles can have a significant effect on profitability.
If demand strengthens while supply remains disciplined, NAND manufacturers could benefit from improved pricing and margins.
However, if manufacturers increase production too aggressively, additional supply could eventually put pressure on prices.
SanDisk’s results and guidance may help investors assess whether improving memory conditions are translating into stronger demand and pricing across the NAND market.
The important distinction is that strong HBM demand does not automatically guarantee equally strong NAND performance. The two markets have different supply conditions and product economics.
4. Teradyne: The equipment side of the semiconductor cycle
Semiconductor manufacturers cannot simply produce increasingly complex chips without extensive testing.
Teradyne supplies automated test equipment used to check semiconductor performance and reliability.
As chip designs become more sophisticated, testing requirements can become more demanding. This creates opportunities for equipment suppliers as semiconductor manufacturers invest in new production capacity and advanced technologies.
Teradyne’s results therefore offer another perspective on the semiconductor cycle.
However, equipment demand can be uneven because customers do not necessarily increase capital spending at a consistent rate. Orders may depend on production schedules, capacity utilisation and the timing of new chip programmes.
For investors, the question is whether demand from advanced semiconductors and AI-related applications can support sustained growth rather than a temporary improvement in orders.
5. Which part of the AI supply chain has the strongest earnings support?
The three main areas offer different opportunities and risks.
HBM memory: SK Hynix offers direct exposure to specialised memory used in advanced AI accelerators. Its technological position and expanding HBM shipments are important advantages, but its exceptional margins and valuation leave little room for disappointment if expectations are too high.
NAND and enterprise SSDs: SanDisk provides exposure to flash storage demand. The potential upside depends on pricing discipline, supply conditions and whether enterprise demand can support a sustained improvement in profitability.
High-capacity hard drives: Seagate benefits from the growing need to store vast quantities of data. Its strong free cash flow is a useful measure of financial performance, although investors still need to assess future demand, pricing and the sustainability of current margins.
These are not interchangeable investments. HBM is closely connected to advanced AI computing, NAND is sensitive to flash-memory supply and pricing, and hard drives serve a broader storage market.
The strongest recent results do not necessarily identify the best-performing stock from here. Valuation, future guidance and the expectations already reflected in share prices matter just as much.
6. What could derail the memory rally?
There are several risks worth monitoring.
First, memory manufacturers could increase capacity in response to attractive margins. If supply eventually grows faster than demand, pricing power could weaken.
Second, major technology companies could reassess their AI infrastructure budgets if the commercial returns from AI applications fall short of expectations.
Third, investors may already be pricing in substantial future earnings growth. Even if companies report strong results, their shares could decline if guidance fails to exceed market expectations.
Finally, the industry remains exposed to changing product demand, technological competition and broader economic conditions.
These risks do not mean the AI memory cycle has peaked. They simply mean that investors should distinguish between a strong industry outlook and an attractive entry price.
My take
The most interesting development is that AI demand is creating opportunities across the entire infrastructure chain, from specialised HBM to enterprise SSDs, high-capacity hard drives and semiconductor testing equipment.
SK Hynix demonstrates the potential profitability of leading-edge memory. Seagate highlights the importance of storage and cash generation, while SanDisk’s upcoming report could help clarify the strength of the NAND recovery.
I would pay particular attention to forward guidance, pricing trends, capital expenditure and free cash flow rather than extrapolating the latest quarterly margins.
The memory rally can continue if AI demand remains strong and supply stays disciplined. But the next phase may depend less on how much infrastructure companies build and more on whether demand can sustain today’s exceptional profitability.
For discussion purposes only
Comments