Author: Li Jia, Wall Street Journal

AI is entering a new stage from "can answer" to "can execute."

According to Chase Wind Trading Platform, Goldman Sachs' latest report shows that the commercialization of AI is shifting from "subscription by seat" to charging based on consumption, transaction volume, and outcomes; meanwhile, Agents are transitioning from auxiliary tools to workflow executors, and the industrial value is migrating from the model itself to proprietary data, business context, and domain expertise.

This means that competition in the AI industry is shifting from "whose model is stronger" to "who can truly master workflows." Model capabilities remain important, but whether they can enter enterprise production environments, understand business contexts, and reliably complete tasks will become a more critical competitive barrier.

This judgment is based on Goldman Sachs' recent on-site investigation of the AI industrial chain in Silicon Valley. From August 18 to 19, Goldman Sachs visited AI startups, leading venture capital firms, and researchers from Stanford University, the University of California, Berkeley, and the University of San Francisco for the third consecutive year. Goldman Sachs believes that as Agents accelerate deployment, the value distribution among cutting-edge models, open-source models, world models, enterprise software, and proprietary data will change.

Agent Deployment: What Enterprises Truly Lack is Not Capability, but "Controllability"

If in the past, AI solved the issue of "helping people complete tasks," then Agents are attempting to solve the issue of "completing tasks by themselves." However, during the large-scale deployment process in enterprises, the biggest obstacle may no longer be model capability, but how responsibilities are divided and whether the entire execution process can be controlled.

The report cites Stanford researchers who point out that most enterprises are still in a manual supervision mode. Especially in fields such as law, risk control, insurance, and auditing, once a model makes a mistake, who is responsible, how to trace the process, and whether it can be corrected in time may be equally as important as the model's capabilities.

Therefore, the workflows that are most easily automated typically have three characteristics: clear decision boundaries, verifiable results, and errors that can be rolled back. Invoice processing is a typical case. AI is responsible for extracting fields and performing checks, low-confidence cases are reviewed by humans, and then the entries are completed through a reversible ERP process.

This also means that information service providers with trustworthy content, verified domain models, and established regulatory relationships are more likely to enter enterprise production environments first.

Model Competition: Division of Labor Between Cutting-Edge Models and Open-Source Models

Regarding the debate of "open source or closed source," the signals released during Goldman Sachs' examination suggest not a binary choice but that different models may correspond to different levels of workflows.

The cutting-edge model camp believes that enterprise benchmark testing often underestimates model capabilities. In real production environments, the business losses caused by model accuracy decline may far exceed the savings from reduced inference costs. Therefore, although many AI-native companies claim to adopt a multi-model strategy, they still heavily rely on cutting-edge models in core production environments.

Another viewpoint holds that the vast majority of enterprise workflows do not require cutting-edge intelligence. As the performance of open-source models continues to improve, customers are increasingly willing to exchange limited performance loss for lower inference costs. A venture capital firm predicts that over the next 12 to 18 months, approximately 90% of inference tokens will flow to open-source models.

This implies that in the future, the AI model market may form a more defined division of labor: cutting-edge models will handle high-value, highly reliable complex tasks, while open-source models will undertake larger-scale standardized tasks and the bulk of token consumption.

World Models: AI Computing Power May Welcome a Second Growth Curve

Over the past 18 months, researchers have increasingly shifted their focus from LLM to "world models."

Unlike LLMs that mainly rely on internet data for training, world models need to understand environments, causality, physical laws, and dynamic interactions in the real world, with data coming more from physical systems, specific industries, and real operational scenarios. This means that the importance of proprietary data may further increase.

Goldman Sachs believes that the problem space corresponding to areas such as physics, industry, science, and robotics is far greater than that of pure text generation, and these workflows often require higher computing power investment. As AI further enters the physical world from the digital world, the demand for computing power in model training, simulation, and inference may experience a new growth curve.

Goldman Sachs predicts that over the next five years, computing power demand may grow by approximately 24 times, and supply-demand tensions are expected to persist for a longer duration, benefiting cloud computing and computing infrastructure companies such as Microsoft, Oracle, and CoreWeave.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。