Facing enterprise software giants: OpenAI seeks

# Dong Jing

At an investor meeting, OpenAI explicitly named giants including Salesforce and Workday as replacement targets, openly declaring its disruptive intentions. Meanwhile, Anthropic emphasized collaboration with existing software through Claude Cowork, positioning itself as “replacing human labor rather than replacing software,” triggering a broad rebound in enterprise software stocks.

OpenAI and Anthropic are taking sharply contrasting strategic stances toward the enterprise software market — one openly vowing disruption, the other deliberately emphasizing partnership. Analysts believe this divergence is reshaping market expectations around AI’s impact on the traditional software industry.

On February 24, Anthropic unveiled details of new features for its Claude Cowork AI software, demonstrating how enterprises can use the tool to access and retrieve data stored in business applications such as DocuSign, LegalZoom, and Salesforce.

Following the announcement, enterprise software stocks that had been under pressure from AI disruption expectations rebounded broadly: Figma rose 10%, ServiceNow edged up 1.4%, and Salesforce gained 4%. The positive market reaction showed that Anthropic’s actual positioning is less aggressive than some investors had feared.

Meanwhile, at an investor meeting last week, OpenAI stated clearly that its AI agents and future products would be capable of replacing software from tech companies including Salesforce, Workday, Adobe, and Atlassian. It presented revenue figures for these enterprise software firms alongside OpenAI’s own 2030 revenue forecasts to investors.

Analysts noted that this private stance stands in stark contrast to Anthropic’s public messaging and has further heightened defensive sentiment across the enterprise software industry.

## Anthropic bets on “replacing labor,” not software

Technology outlet The Information reported that, when launching new Claude Cowork features, Anthropic deliberately centered its narrative on replacing human labor rather than displacing software tools. Its logic:

The new AI tools will continue to call on existing enterprise software, meaning companies will still pay for these applications, and revenue streams for software vendors will not disappear.

Anthropic invited its Chief Economist Peter McCrory to the launch event. McCrory stated that AI’s impact on the labor market would be “extremely uneven”: high-skilled workers would use AI to boost productivity, while lower-skilled workers in basic data-entry roles faced replacement risks.

Derek Hernandez, an analyst at PitchBook Data, added that Anthropic’s Claude features targeting financial services could pressure jobs in the sector, saying, “It’s investment banking seats, equity research seats, especially entry-level white-collar roles that will be hit.”

Notably, the report pointed out that Anthropic employees still use a wide range of traditional enterprise software applications internally. While this detail does not fully eliminate concerns among software vendors — whose business models depend on massive enterprise user bases — it sends a distinctly different market signal compared to OpenAI.

## OpenAI plays its “disruption card” to investors

By contrast, OpenAI’s ambitions in the enterprise software market are more overt.

Reports said OpenAI explicitly named enterprise software giants including Salesforce, Workday, Adobe, and Atlassian as potential replacement targets at last week’s investor meeting. It showed investors the revenue scale of these companies versus OpenAI’s 2030 revenue projections, aiming to convey the size of its addressable market.

OpenAI also shared a calculation with investors:

The average employee using ChatGPT saves roughly 50 minutes of work time per day, equivalent to about $50 per person per day. The enterprise subscription for ChatGPT starts at just $25 per employee per month. On this basis, OpenAI argued it currently captures only a tiny fraction of the value it creates. The estimate partly references analysis from OpenAI shareholder Ark Invest.

These private comments align with widespread interpretations of OpenAI’s new “Frontier” AI product launch the previous month: OpenAI is seeking to position its technology as an overarching gateway for enterprise applications, gradually influencing corporate purchasing decisions for software and AI by controlling how business data is accessed.

## Traditional software vendors build defenses; customer behavior shifts quietly

Facing the advance of AI labs, traditional enterprise software vendors are actively building defenses.

Firms including ServiceNow and Microsoft are emphasizing to customers that their software offers superior reliability and compliance compared to experimental products from AI labs. HubSpot is considering charging extra fees to customers who want AI agents to access data within its systems.

Yet it remains unclear whether traditional vendors can truly block the infiltration of AI agents. The core design of AI agents is precisely to take over user devices and operate various applications in a human-like manner.

Analysts noted that the strongest current moat for traditional enterprise software firms lies in their long-established global expertise in data compliance and privacy regulation — capabilities experimental AI products cannot easily replicate in the short term.

Although most enterprise clients have not yet genuinely attempted to replace existing enterprise software with AI, AI agents are changing how employees interact with software and creating real cost-replacement effects in certain scenarios.

According to reports, a cybersecurity executive said he avoided over $100,000 in annual subscription fees for CrowdStrike software by connecting to an AI agent. The CrowdStrike product had been used to automate employee account management, including flagging suspicious logins and identifying and locking dormant accounts for employees suspected of leaving.

The executive instead adopted an AI agent from startup Torq, powered by OpenAI and Anthropic models, which directly uses raw login data already collected by the company via Microsoft software to deliver equivalent functionality at a lower cost.

This case reveals the potential impact path of AI on the enterprise software market: even if enterprises do not fully abandon existing software in the short term, AI agents could make some software products “dispensable” in employees’ eyes and gradually establish AI itself as the core tool for work. CrowdStrike said it has allowed AI agents to integrate with its software so customers can use both together.

---

## Risk Warning and Disclaimer

The market involves risks, and investments require caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are appropriate for their particular circumstances. Investment decisions based on this article are made at the user’s own risk.