What we know
A recent report indicates that the cost of AI "intelligence"—defined as the expense involved in training and running AI models—has been declining at a rate faster than Moore's Law. Moore's Law traditionally describes the doubling of computing power approximately every two years. This trend was highlighted by Tom's Hardware, which noted that the price of AI capabilities has fallen dramatically in recent years, potentially faster than any other transformational technology in history. However, this claim remains unverified due to a lack of independent confirmation from multiple authoritative sources.
Why it matters
This topic is being examined as an explainer on the TECHNOLOGY desk. The Intel Brief is not endorsing vendor claims as fact and advises caution in interpreting language such as "smarter" or "stronger," which reflect the source’s framing rather than confirmed findings. Moore's Law has historically served as a key benchmark for technological progress, forecasting exponential improvements in computing power alongside cost reductions. The reported rapid decrease in AI training and inference costs, if accurate, could signify a significant acceleration in technological advancement. Nevertheless, these claims require thorough, independent validation before they can be accepted as established fact, especially given their potential implications for the AI industry and broader technology landscape.
What is still unknown
The current information is based on fewer than two independent sources, meaning the claim that AI costs are falling faster than Moore's Law is not independently verified. The Intel Brief has not conducted any independent testing of the products, patches, or attacks referenced. As a result, any details regarding the technical effects, timelines, or customer impacts beyond what is included in the report remain unknown. Further investigation and corroboration are necessary to confirm the accuracy and significance of these claims.