Matthew Berman Predicts AI Costs Will Continue to Decline
Matthew Berman stated that AI costs will decrease, token efficiency will improve, and AI outputs will be more effectively disseminated.
Open-source models will further drive down prices, and enterprises' ability to adopt AI will also strengthen, currently in a "strange mid-stage."
Market Mechanism: Developers and enterprises as buyers accelerate AI procurement, capital flows from high-cost closed-source models to efficient open-source and optimization tools, AI infrastructure suppliers face short-term pressure but will benefit in the long term from increased penetration, with thought leaders like Matthew Berman reinforcing market optimism through their views.
Supplementary Data: Berman has recorded a related video detailing this viewpoint.
Source: Public Information
ABAB AI Insight
Matthew Berman has long been an independent AI analyst, having accurately predicted improvements in model performance and declines in cost multiple times. This viewpoint continues his tracking of the maturity path of AI technology, having previously pointed out the transition from "expensive and scarce" to "affordable and widespread."
On the capital path, Berman emphasizes that enterprises are reducing the cost per token through internal optimization and migration to open-source, motivated by the shift from the current high-investment validation phase to large-scale deployment to achieve a positive ROI cycle, while also pushing the entire industry from experimental spending to productivity tools.
Similar to the past process of cloud computing from early high prices at AWS to today's significant cost reductions, the AI industry is currently in a "strange mid-stage" transitioning from frontier exploration to widespread commercial adoption, focusing on solving efficiency bottlenecks and organizational adaptation issues.
Structural Judgment: This essentially belongs to technological substitution. Open-source models and efficiency improvements are replacing high-priced closed-source large models, achieving the transition of AI from exclusive tools for a few giants to a public productivity infrastructure through cost reduction and accelerated diffusion. The mechanism is that the rapid improvement of token economics breaks down adoption barriers, accelerating penetration across the industry.
ABAB News · Law of Cognition
The highest costs always appear on the eve of technology popularization.
The strange mid-stage is the hardest to endure but the most valuable.
Efficiency improvements reshape the industry landscape more than increasing parameters.