Indxx reports on AI-related semiconductor growth

Indxx researchers with Alan J Weissberger presented an overview of forecasts and developments for AI-related semiconductors and noted implications for hardware deployment and cost planning.

Multiple research firms and banks forecasted that revenue from AI-related semiconductors would grow at about 18% annually. IDC forecast that global Artificial Intelligence (AI) hardware spending, including chip demand, would grow at 18%; Morgan Stanley, Infosys, MarketResearch.biz, IEEE IRDS, and Citi issued similar 18% growth projections for various AI hardware and chip categories.

The report described purpose-built AI processors — including ASICs, FPGAs, NPUs, and Google’s TPUs — as designs optimized for neural networks, training, and inference workloads; it said these chips were optimized for matrix multiplications and convolutions and delivered higher performance-per-watt than CPUs or GPUs as Large Language Model (LLM) and Generative AI (GenAI) workloads expanded.

The report summarized a multi-year, multi-billion-dollar agreement announced in October 2025 in which OpenAI would design hardware and Broadcom would develop custom chips and supporting infrastructure using Ethernet networking; the arrangement covered 10 gigawatts of AI compute with deployments expected over four years and potentially extending to 2029 and included projected cost savings of 30–40% versus off-the-shelf Nvidia or AMD chips.

It also noted that Google announced Ironwood, its seventh-generation Tensor Processing Unit (TPU), which offered four times the performance of the prior Trillium generation and supported superpods of up to 9,216 interconnected chips for large LLM training and low-latency inference. The report cited estimates that NVIDIA held about an 86% share of the AI Graphics Processing Unit (GPU) segment per one source, with other estimates ranging 80%–92%, while AMD’s discrete GPU share was placed at roughly 4%–7% and Intel near 1%.

The report recorded additional projections that edge AI chips would reach $13.5 billion in 2025, AI accelerators based on Application-Specific Integrated Circuit (ASIC) designs would grow about 34% year-over-year in 2025, automotive AI chips would surpass $6.3 billion in 2025, and U.S.-based AI chip startups raised over $5.1 billion in venture capital in the first half of 2025, and it concluded that custom silicon was now essential for deploying AI in real-world applications such as automation, robotics, healthcare, finance, and mobility.