Defiance ETFs has launched the Defiance Inference AI Chip ETF (NASDAQ:AINF), giving investors targeted exposure to the chips powering the next stage of the artificial intelligence buildout: inference.
AINF tracks the BITA AI Inference Chip Select Index, which focuses on global companies involved in designing, fabricating or manufacturing specialized inference processors, including GPUs, ASICs and neuromorphic chips. The strategy spans both data center accelerators and lower-power chips designed for edge AI applications. The fund carries an expense ratio of 0.65%.
The launch comes as AI investment shifts from training models toward running them at scale. Every chatbot response, AI agent task and deployed AI application generates inference workloads, potentially creating a recurring source of chip demand as AI adoption expands.
QUICK CONTEXT: AI Chip Demand Enters Next Phase
Much of the AI investment boom has centered on the massive computing infrastructure required to train increasingly sophisticated models. Nvidia and other semiconductor companies have benefited from demand for high-performance GPUs and accelerators used in hyperscale data centers.
Inference represents the next stage of that computing cycle. Unlike training, which involves building or updating an AI model, inference occurs every time a trained model generates an output. That can include a chatbot answering a question, an AI assistant completing a task or an autonomous system processing new information.
The distinction matters for investors because inference workloads can scale with AI usage rather than simply with the creation of new models. As AI applications move into enterprise software, smartphones, vehicles and edge devices, demand may increasingly favor chips optimized for speed, power efficiency and cost.
AINF gives investors a targeted vehicle for that theme, focusing specifically on companies developing the semiconductor technologies needed to execute AI models rather than offering broad exposure to the entire AI ecosystem.
Photo: Shutterstock
Login to comment