March 2026

Ilaria Sangalli, Head of Index Insights, EMEA

Technology equity performance diverged meaningfully during February 2026, as market action reflected a more granular reassessment of AI exposure rather than a broad‑based de‑risking of the sector.

Software stocks came under pressure as investors weighed the risk that generative AI and agentic tools could structurally compress pricing power and lower barriers to entry across parts of the application layer.

That concern was not evenly expressed across technology. Semiconductor and infrastructure‑oriented companies proved comparatively resilient, supported by continued visibility into AI‑driven demand for compute, memory, and networking capacity.

Dispersion within software also widened. Cybersecurity names held up better than many application‑software peers, benefiting from mission‑critical use cases and more defensible pricing dynamics, while other software segments experienced more pronounced multiple compression.

This divergence highlights an important distinction for investors: AI value creation is not evenly distributed across the technology stack. While parts of application software face rising commoditisation risk, companies positioned around hardware, memory, edge intelligence, and networking infrastructure continue to benefit from AI’s rapid diffusion.

Recent market behaviour provides evidence of this distinction. This article spotlights four constituents of the Nasdaq Global AI and Big Data^TM Index (NYGBIG^TM) that are not included in the Nasdaq‑100® (NDX®) or S&P 500 (SPX) and demonstrated relative resilience during February’s software‑driven sell‑off.

Each plays a structurally important role in the AI ecosystem and holds a significant position in the global AI patent landscape.
 

Source: Bloomberg. Data as of February 27, 2026

STMicroelectronics: Intelligence at the Edge

STMicroelectronics (STM), a European semiconductor company (index weight of 0.31%)¹ operates where software meets the physical world.

The company specialises in microcontrollers, sensors, and power semiconductors embedded in vehicles, industrial systems, energy infrastructure, and consumer devices, where workloads operate under strict constraints on latency, reliability, and power consumption.² This positioning allows STM to benefit from rising silicon content per device, as more compute, sensing, and control are embedded directly into physical systems.

During the February software sell‑off, STM delivered a +18.78% USD total return³, reflecting investor recognition that edge intelligence is becoming more valuable as AI workloads decentralise.

STM’s patent portfolio reinforces this positioning. The company reports approximately 21,000 active and pending patents,⁴ underpinned by substantial R&D investment of around $2bn in 2025 (~17% of revenue).⁵

Taking recent patent grants as illustrative examples, its innovation spans across:

• AI‑enabled MEMS⁶ and inertial navigation - enabling accurate localisation without GPS⁷
• On‑sensor machine learning - where decision‑tree‑based inference runs directly inside ultra‑low‑power sensors⁸
• 3D sensing and biometric security - supporting “wake‑on‑attention” and low‑latency perception⁹

Product execution mirrors this strategy. STM has introduced Stellar P3E, one of the first automotive microcontrollers with a built‑in AI accelerator, targeting software‑defined vehicles (SDVs). By moving AI processing from centralized vehicle computers to the vehicle edge, Stellar P3E enables sub‑millisecond decision‑making with lower latency, power, and thermal overhead - key requirements for next‑generation automotive architectures.¹⁰

SK hynix 

SK hynix (index weight of 5.02%) is a South Korean memory semiconductor manufacturer specialising in DRAM- including high‑bandwidth memory (HBM) - and NAND flash, and is a key supplier to AI chipmakers such as Nvidia.¹¹,¹²

As AI workloads have scaled, memory has increasingly emerged as the system‑level bottleneck, elevating the strategic importance of HBM and advanced DRAM. Memory capacity and bandwidth now directly constrain model throughput, achievable context length, and end‑user latency, effectively turning memory into a visible product feature rather than a hidden infrastructure input.¹³

Against this backdrop, SK hynix has established a leading position in high‑bandwidth memory for AI chipsets, commanding an estimated 61% share of the global HBM market, according to Macquarie Equity Research.¹⁴ The company’s HBM revenue more than doubled in 2025, reflecting strong demand from AI accelerators and data‑centre deployments.¹⁵

This structural importance has begun to be recognised by investors. During a February sell‑off in software equities, SK hynix delivered a +6.27% USD total return, underscoring its role as a critical beneficiary within the AI hardware value chain.¹⁶

SK hynix’s leadership in AI memory is supported not only by scale and execution, but also by the company’s sustained focus on innovation. A recent example is a patent granted in February 2026 (“Three‑Dimensional Semiconductor Device and Method of Manufacturing the Same”) which addresses critical challenges in scaling 3D memory for AI workloads. This patent relates to a 3D stacked memory architecture in which multiple memory layers are vertically integrated rather than laid out horizontally. It introduces the use of different insulating materials above and below each conductive layer to reduce electrical interference, structural damage, and manufacturing defects as layer counts increase.¹⁷

Kingsoft Cloud  

Kingsoft Cloud (index weight of 0.06%) is one of China’s leading independent cloud computing providers, providing cloud storage and computation services.

Over the past two years, the company has strategically pivoted toward its Intelligent Computing Cloud, which has become its primary growth engine by supporting AI model training and inference.¹⁸

Kingsoft Cloud also plays a central role in China’s AI ecosystem as the primary AI compute and cloud partner for Xiaomi’s large language model initiatives and WPS AI.¹⁹ This AI‑led positioning was further underscored at the July 2025 World Artificial Intelligence Conference, where the company showcased a suite of full‑stack, industry‑focused AI solutions, including:²⁰

• StarFlow Platform – a cloud‑native platform for AI model training and inference. It was launched to meet the demands of the general trend of transitioning from large customers' highly concentrated training demand to more widespread customers' inference demand. Focused on cost efficiency, flexibility, and end‑to‑end lifecycle management for enterprises and developers
• Hanhai Big Model Development & Application Platform: an industry‑centric platform. Practical AI use cases rather than generic models. The platform lowers technical barriers for enterprises and supports closed‑loop deployment, particularly in regulated and data‑sensitive sectors such as government, legal services, finance, and healthcare.
• The Government AI All‑in‑One Machine - integrates domestic computing hardware with private AI models. This approach addresses key public‑sector requirements such as data sovereignty, security, and system control, enabling government clients to adopt AI within compliant, self‑contained environments.

This growing exposure to AI compute demand reinforced investor confidence and helped the stock hold up during the broader software sector sell‑off. Between February 2 to February 13, 2026, the Kingsoft Cloud ADR delivered a 6.24% total return in USD.²¹

Nokia Oyj

Nokia Oyj (index weight of 0.64%) is a global provider of mobile, fixed and cloud network solutions, combining hardware, software, services, and intellectual property licensing. It operates through the following segments: Mobile Networks, Network Infrastructure, Cloud and Network Services, and Nokia Technologies.²²

After a difficult 2024 marked by declining revenue, Nokia entered 2025 in turnaround mode, catalysed by the appointment of a new CEO, Justin Hotard, and a strategic pivot toward AI‑enabled networking and next‑generation wireless infrastructure. This repositioning, combined with early signs of operational improvement, helped drive a strong rebound in the stock, which rose 46% over the course of 2025.²³ One of the most significant catalysts behind this renewed optimism is Nokia’s strategic partnership with Nvidia. Together, the two companies are developing AI‑driven radio access networks that enhance 5G‑Advanced capabilities while laying the groundwork for AI‑native 6G.²⁴

This exposure to long‑cycle AI infrastructure, rather than discretionary software spending, has helped insulate Nokia from the recent sector sell‑off, with its ADR delivering a 6.12% total return in USD between February 2 and February 13, 2026.²⁵

This resilience also reflects Nokia’s renewed focus on innovation. At the beginning of 2025, Nokia announced that it had reached a milestone of 7,000 patent families declared as essential to the 5G standard, underscoring its leading role in cellular technology innovation. These patents cover foundational technologies that define how smartphones, connected vehicles, IoT devices, and other connected products interact with 5G networks, including advances in radio protocol design, security, and interface technologies.²⁶

Nokia is consistently ranked among the top global owners of 5G Standard Essential Patents (SEPs) in independent industry analyses such as the Patently 100 – 5G 2025. It sits in the top tier worldwide, alongside companies like Huawei, Qualcomm, Samsung, Ericsson, and LG.²⁷ This position is underpinned by more than $100bn invested in R&D since 2000, with over $5bn invested in 2025 alone (~23% of sales).²⁸