Semiconductor Capex Set to Hit $3.2T by 2030

Bank of America projects the global semiconductor industry will expand from $1.7 trillion in 2024 to $3.2 trillion by 2030, representing an 88% increase. Analyst Vivek Arya’s note indicates that despite market concerns regarding slowing artificial intelligence investment, order volumes, capacity commitments, and pricing remain firm. The firm attributes this sustained momentum to robust demand across memory, core logic, and server infrastructure.

The forecast highlights memory as the largest growth segment, expected to rise 92% to $1.8 trillion. Core semiconductor sales are projected to grow 83% to $1.35 trillion, while server sales are forecast to surge 136% to $848 billion. This structural shift positions specific manufacturers in a favorable position to capture value from the anticipated surge in hardware spending over the next five years.

Memory Sector Leads Growth

SK Hynix is positioned to benefit from memory becoming the largest market segment, which will see an absolute dollar increase of $863 billion. The company leads in high bandwidth memory, a specialized product packaged with graphics processing units to reduce latency in AI workloads. While high bandwidth memory commands premium margins, standard dynamic random access memory and flash prices have also seen significant increases.

As the overall dynamic random access memory market moves toward balance, SK Hynix’s revenue mix, which is heavily weighted toward high bandwidth memory, offers a strategic advantage over competitors like Micron. This exposure to high-margin products allows the firm to capitalize on the specialized demand driving the broader memory surge, making it a primary beneficiary of the forecasted expansion.

Nvidia Dominates Compute Infrastructure

Nvidia maintains a dominant position in the compute sector, where its graphics processing units remain the standard for training artificial intelligence models. The company’s software ecosystem, built on the CUDA platform, creates a significant barrier to entry, as most early AI code is written specifically for its architecture. This lock-in effect secures Nvidia’s role in the foundational layers of AI development.

The firm is expanding its influence in the inference market by integrating technology from Groq. This allows Nvidia to utilize language processing units for memory-intensive decode phases and graphics processing units for compute-heavy pre-fill phases. Combined with its networking portfolio, this end-to-end server approach addresses the full spectrum of AI tasks, from training to agentic workflows.

ASML Secures Lithography Monopoly

Bank of America forecasts wafer fabrication equipment spending will rise 129% from $156 billion to $360 billion by 2030. ASML is the primary beneficiary of this capital expenditure, holding a monopoly on extreme ultraviolet lithography technology. This technology is essential for manufacturing advanced logic chips and high bandwidth memory, making ASML a critical supplier for the industry's most complex components.

The company plans to increase its extreme ultraviolet capacity by 30% next year and by another 30% in 2028. Its newest high numerical aperture machines, which command double the price of standard extreme ultraviolet units, have secured commitments from major foundries. This capacity expansion aligns with soaring demand for graphics processing units, central processing units, and memory, ensuring sustained revenue growth for the Dutch firm.