The global technology and semiconductor industry is entering a structurally stronger but increasingly uneven growth phase, driven by artificial intelligence (AI), high-performance computing (HPC) and next-generation connectivity. However, analysts warn that intensifying memory shortages, rising input costs and geopolitical tensions are beginning to distort the traditional semiconductor upcycle, creating widening divergence across end markets.

At the center of this shift is the rapid rise in demand for generative AI and data center infrastructure, which is absorbing an increasing share of global memory supply. According to MBSB Research, this has triggered a structural reallocation of memory chips away from consumer electronics such as smartphones, PCs and other devices toward AI servers and high-performance computing applications.

The result, it said, is tighter availability of memory for smartphones, leading to higher input costs, constrained device supply and elevated selling prices. Smartphone manufacturers are increasingly expected to respond by adjusting either pricing or device specifications, which could weigh on demand, particularly in the low-to-mid-end segment.

In contrast, the premium segment is expected to remain relatively resilient due to stronger margins and more stable supply chain positioning.

MBSB Research noted that this dynamic is already reflected in a weaker outlook for the smartphone industry. Global smartphone production in 2026 is now expected to decline marginally year-on-year, although more recent estimates point to a deeper contraction.

IDC is projecting a 12.9 percent decline in global smartphone shipments in 2026, while Counterpoint estimates a 12 percent fall. If realized, this would mark the lowest annual smartphone shipment level in more than a decade.

The research house added that while there are ongoing efforts to expand memory capacity, particularly from suppliers in China, these additions are unlikely to be sufficient to offset the surge in AI-driven demand. As a result, memory tightness could extend beyond 2026, suggesting a more persistent structural constraint rather than a cyclical imbalance.

Kenanga Research similarly highlighted that AI workloads are significantly more memory-intensive than traditional computing applications, leading to prioritization of high-bandwidth memory (HBM) and advanced DRAM for AI servers, while reducing available supply for smartphones, PCs and automotive electronics.

It described this as an increasingly “zero-sum” environment in the memory segment, where hyperscale cloud operators are absorbing a disproportionate share of output, resulting in tighter supply and weaker affordability in non-AI end markets.

Against this backdrop, Apex Securities said structural demand recovery across the technology sector is expected to broaden in 2026, underpinned by continued investment in AI infrastructure, optical connectivity, advanced packaging and semiconductor back-end equipment.

It noted that commentary from industry engagements and site visits points to several consistent trends, including healthy order pipelines, ongoing capacity expansion and improving conversion of qualification cycles into revenue.

The firm highlighted that AI-related demand continues to flow through multiple layers of the semiconductor supply chain, supporting a broader recovery across equipment, components and enabling technologies.

However, Apex also indicated that growth is increasingly being driven by AI-linked segments, while non-AI end markets remain comparatively uneven, reflecting the same demand concentration highlighted by other research houses.

Despite these pressures, Kenanga Research said the global semiconductor cycle remains firmly intact, with the industry still on track to approach $1 trillion in value by 2026. The sector continues to be supported by structural drivers including AI, HPC, 5G deployment and next-generation device upgrades.

However, it said growth is becoming increasingly uneven as AI-driven demand concentrates supply in specific segments, particularly memory. This is extending the cycle but also amplifying divergence across end markets.

Kenanga added that AI applications are evolving from assistants and copilots toward more autonomous “agentic AI” systems, where software can execute tasks with limited human intervention.

This transition is expected to extend the investment cycle beyond model training into broader deployment, sustaining long-term demand for AI infrastructure. At the same time, the focus is shifting toward monetization, as industry players seek to convert AI capabilities into revenue-generating applications.

On the supply side, Kenanga said memory is expected to lead the next phase of wafer fabrication equipment (WFE) investment, driven by highly equipment-intensive migration in DRAM and 3D NAND technologies. Memory is expected to account for a dominant share of total WFE capital expenditure over the coming years, reinforcing its central role in the semiconductor investment cycle.

While structural drivers remain supportive, analysts highlighted that external risks are increasingly clouding the outlook. Kenanga pointed to rising geopolitical tensions, particularly in the Middle East, as a potential source of disruption to global technology supply chains and AI infrastructure expansion.

MBSB Research also flagged geopolitical risks, noting that tensions in the Middle East could affect global technology firms expanding AI infrastructure in the region. It highlighted growing exposure of US technology companies, which have been increasing investments in data centers and cloud infrastructure across the Middle East.

The research house added that disruptions to digital infrastructure in the region in recent years underscore vulnerabilities in increasingly interconnected cloud and AI ecosystems, where physical, digital and geopolitical risks are becoming more closely linked.

Overall, analysts said the semiconductor and broader technology sector continues to be supported by strong structural drivers, but is increasingly characterized by uneven growth, concentrated AI-led demand and rising external risks.

While the AI supercycle remains intact, intensifying memory constraints and geopolitical uncertainty are reshaping the industry into a more fragmented and risk-sensitive growth phase.

Analysts see data center to boost Malaysia’s construction growth