AI applications are entering a critical next phase—shifting from developer tools to a comprehensive penetration into knowledge workers. Goldman Sachs' latest research report outlines a clear path for the scaling of general intelligence tools through two iconic products—Claude Cowork and OpenClaw.
On March 7th, according to news from the Chasing Wind Trading Desk, Goldman Sachs pointed out in its latest research report that the next phase of AI as a general intelligence layer application has emerged, with Anthropic's Claude Cowork and the open-source agent OpenClaw being two core data points defining this trend.
The report states that Claude Cowork, as the first &#34;trusted case of agent workflow&#34; aimed at knowledge workers, is reshaping the value distribution in the SaaS industry—it will not directly replace existing record systems (such as CRM or ERP), but will shift market share towards SaaS providers that can effectively integrate AI capabilities.
Meanwhile, OpenClaw's viral spread not only showcases a new form of human-computer interaction for the future, but its unique hosted economics are breaking the monopoly of hyperscalers, directing the massive new demand for reasoning computing power directly to alternative edge computing and VPS (Virtual Private Server) providers like DigitalOcean and Cloudflare.
<h2>Claude Cowork: The First Trusted Case of Agent Workflow for Knowledge Workers</h2>
On January 12th, Anthropic officially launched Claude Cowork—a proxy tool embedded in the Claude desktop application, designed for non-developers, capable of automating file management and task processes, covering multi-step operations such as file organization, document processing, and workflow management.
On January 30th, Anthropic further released a series of Cowork plugins aimed at &#34;business functions,&#34; covering scenarios such as sales and marketing, finance, legal, customer support, data analysis, productivity and enterprise search, as well as product management.
Goldman Sachs believes that the core value of Cowork lies in filling the critical gap between developer-specific proxy tools and the chatbots commonly used by knowledge workers. Software engineers have been gaining productivity through terminal proxy loops for over a year, while AI interactions for knowledge workers have long remained at the chat interface level. Cowork extends this capability to non-technical personnel through an application interface integrated with Excel and standard business toolsets.
Specific implementation scenarios include: accountants categorizing credit card transactions, HR teams compiling tax reports and tracking PTO, and CFOs accelerating project-specific DCF analyses. Among digital agent clients, marketing scenarios are the most prominent—clients are integrating Klaviyo and automating the entire reporting process.
Cowork's core competitive advantage lies in its role as an intelligent layer across the SaaS stack: it does not depend on any single SaaS product, can switch freely between tools, and can reason over data that spans multiple systems and &#34;glue layers.&#34;
Goldman Sachs' industry research shows that Anthropic is clearly leading in both tool capabilities and underlying model intelligence.
<h2>Cowork will not replace system records but will reshape SaaS value distribution</h2>
Goldman Sachs clearly states that tools like Cowork will not replace core system records such as CRM and ERP, and companies have no intention of dismantling their existing SaaS stack. However, as AI becomes a key differentiator, wallet share will concentrate among SaaS vendors that can provide the most effective AI workflows.
In practice, customers are using Cowork to replace lightweight &#34;glue&#34; software—tools primarily designed to move data between systems—rather than core platforms. Conversations with Goldman Sachs' partners indicate that companies still prefer to pay third-party software vendors for security, governance, and operational control.
The collaboration case between Intuit and Anthropic is the best annotation of this logic. On February 24, Intuit announced a partnership with Anthropic: Intuit will gain access to Anthropic models and custom AI agents through the Claude Agent SDK, while Anthropic will promote Intuit products through MCP integration on Cowork, Claude for Enterprise, and Claude.ai platforms.
Goldman Sachs believes that the logic of this collaboration is similar to Intuit's previous $100 million annual agreement with OpenAI: the foundational model does not require building custom workflows but monetizes through model capabilities and reach; Intuit retains control over data and maintains its unit economics.
Goldman Sachs notes that the potential risk lies in adding a layer between customers and Intuit, which may abstract value in the long term; however, the potential benefit is better customer acquisition, similar to the collaboration model with search engines during the Web 2.0 era.
It is noteworthy that on March 5, OpenAI launched GPT-5.4—its first general model with native computer usage capabilities, similar to Claude Cowork, capable of operating computers and executing agent workflows across applications. OpenAI particularly emphasizes that this model reduces token consumption by approximately 47% on certain tasks while maintaining the same accuracy as previous models.
<h2>OpenClaw: Breaking context limitations, showcasing new directions for future human-computer interaction and computing power demands</h2>
Goldman Sachs believes that OpenClaw achieves viral growth in consumer applications by granting AI agents complete control over computers and infinite memory; its unique local and low-cost VPS hosting model is becoming a significant boon for edge computing providers.
OpenClaw (formerly known as Moltbot and Clawdbot) was first released in November 2025 and achieved &#34;viral&#34; spread in late January.
<blockquote>
This is an open-source, self-hosted, always-on AI agent that can run on users' local computers, capable of receiving commands through personal messaging platforms like iMessage, WhatsApp, Google Chat, and executing any task that a human can complete, including opening applications, browsing the web, and completing transactions
OpenClaw has solved the key technical bottleneck that restricts most AI agents — the context window problem. Its solution is based on a file-based memory system: all agent activities are stored in Markdown files on the local computer, and the agent loads the day's memory and the previous day's context at the beginning of each session. Users can also establish long-term memories through instructions (e.g., &#34;always prioritize checking OpenTable when booking a restaurant&#34;), forming a persistent preference database.
</blockquote>
Matthew Prince, co-founder and CEO of Cloudflare, pointed out that the development trajectory of OpenClaw over the next three years may replicate the path of ChatGPT over the past three years: it is currently catalyzing a large number of experiments, but similar to the early days of ChatGPT, it may take several years to achieve sustained scaled daily use.
<h2>The Economics of OpenClaw's Computing Power: Structural Opportunities for Edge Computing and VPS Providers</h2>
The hosting economics of OpenClaw provide important insights for investors. Due to the significant cost fluctuations of hyperscale cloud service providers, most users choose to run OpenClaw on home hardware (Mac Mini, old laptops) or low-cost VPS providers (such as Cloudflare and DigitalOcean). Some users have reported token consumption of up to several thousand dollars within the first few days of use.
Many users are adopting a layered model architecture: the frontier model is responsible for orchestration, while smaller models execute sub-tasks, thereby reducing the cost per query.
DigitalOcean's data is the most intuitive: within days of OpenClaw's release, customers deployed nearly 30,000 native one-click OpenClaw Droplets on DigitalOcean, with thousands of additional deployments activated. The share of inference services in DigitalOcean's AI customer revenue has been steadily increasing — by December 2025, the ARR of AI customer revenue is expected to reach $120 million, with non-bare metal business accounting for 70%, and inference services experiencing a year-on-year growth rate of 254%.
Cloudflare has not yet seen a significant impact of OpenClaw on logo growth or network traffic, but Goldman Sachs believes that this trend further confirms the leading cost-performance advantage of Cloudflare's edge network and Workers software

GS-C

GS-A

GS-D

Goldman Sachs believes that AI general intelligence layer applications have entered a new stage. Claude Cowork fills the gap between developer tools and ordinary users, not replacing core systems like CRM/ERP, but reshaping SaaS value distribution. OpenClaw showcases a new form of human-computer interaction, and its low-cost hosting model is breaking the monopoly of large companies, providing new computing power opportunities for edge computing and VPS providers

- Goldman Sachs highlights a pivotal shift in AI applications, transitioning from developer tools to knowledge worker integration exemplified by Claude Cowork and OpenClaw.  
- Claude Cowork is reshaping SaaS value distribution by enhancing workflows for non-developers without replacing existing systems, while OpenClaw's viral adoption demonstrates a new interaction model, leveraging local hosting to meet burgeoning computational demands.  
- Both innovations emphasize the importance of AI in transforming operational efficiencies, with implications for cloud service providers and the tech industry at large.

Wallstreetcn

Goldman Sachs

GOLDMAN SACHS GROUP INC DEP SHS REPSTG 1/1000TH PRF SER C

GOLDMAN SACHS GROUP INC DEP SHR REP 1/1000TH PFD SER A

GOLDMAN SACHS GROUP INC DEP REP 1/1000TH PRF D

Goldman Sachs' judgment: Claude Cowork becomes the first "trusted case for agency workflow," OpenClaw demonstrates the future direction of human-computer interaction