Anthropic is simultaneously pursuing two objectives: bringing itself to Wall Street and securing an external validation of its safety credentials before going public. According to sources familiar with the matter, the AI company's annualized revenue is expected to exceed $100 billion this year, with preparations underway for a listing as early as November. In parallel, Anthropic will embed evaluators from consulting firm Accenture PLC (NYSE: ACN) directly into its operations to conduct safety testing on its latest models.

The first development underscores how rapidly this business is generating revenue, while the second signals the company's intent to resolve the question of "who oversees me" before the market debut. Taken together, Anthropic's IPO readiness can be divided into three pillars: proof of revenue, capital arrangements, and groundwork for reputation and compliance. The first two have clear metrics and timelines; the third is the hardest to quantify and the most likely to face scrutiny.

IPO Timeline: Listing as Early as November, Roadshow Potentially Mid-October

According to reports, Anthropic's revenue run rate—a metric that projects annual revenue based on a shorter period—has surged this year as enterprise clients increasingly adopt its Claude AI software for code simplification and other operational tasks. The reports also noted that alongside its IPO plans, the company is contending with markedly heightened public concern over AI's potential for catastrophic harm. Anthropic declined to comment.

The revenue run rate reached $65 billion at the end of July, up from just $9 billion at the end of 2025. This sharp acceleration in revenue provides substantial support for the company's listing plans. Both Anthropic and OpenAI have confidentially submitted offering documents, with Anthropic expected to complete its Wall Street debut within the coming weeks—marking the start of a race between the two companies for public market capital.

Earlier this month, reports indicated that Anthropic could commence its IPO roadshow as early as mid-October and list just days before the U.S. midterm elections in November. On the scale of the offering, previous reports have offered more specific figures: Anthropic plans to raise up to $100 billion through the IPO at a target valuation of approximately $2 trillion, with Nvidia in talks to participate as a cornerstone investor with an investment of up to $10 billion. The company confidentially filed its S-1 in June and has selected Nasdaq as its listing venue.

It is important to distinguish that the $2 trillion figure represents a market-circulated target valuation, not a final offering price. Social media posts indicate that Anthropic also hosted a venture capital forum this week to discuss progress on its AI work, with multiple investors in attendance. Among the questions put to a public vote at the event was how to determine whether a company has achieved the more advanced form of AI known as "artificial general intelligence" (AGI).

Revenue Run Rate: From $9 Billion to $65 Billion to $100 Billion

The $100 billion figure represents a further extrapolation beyond previous estimates. Anthropic's revenue run rate jumped from $9 billion at the end of 2025 to $65 billion at the end of July; the latest projection places the annualized level at "over $100 billion this year." Anthropic has told investors that second-quarter revenue exceeded $11.5 billion, with an annualized run rate above $65 billion, broadly in line with these figures.

For enterprise customers, Claude's appeal centers on programming efficiency and automation of routine office tasks—the primary driver behind the rising run rate.

Safety Preparations: Accenture Evaluators Embedded, Joint Investment of at Least $1 Billion

In a blog post published Friday, Anthropic stated that evaluators from Accenture PLC (NYSE: ACN) will conduct red-team testing on the lab's latest models, probe its safety mechanisms, and run assessments to ensure the technology is "aligned"—meaning its behavior remains consistent with human objectives. The post noted that these evaluators will have access within the company comparable to that of Anthropic's own employees.

In a 3,800-word essay published Saturday, Anthropic CEO Dario Amodei responded to growing concerns over safety risks in frontier AI models, calling for a slowdown in development pace and announcing new safety measures, with third-party evaluators being one of them. According to earlier public reports, Amodei's rationale for this call is a mechanism known as "recursive self-improvement" that is beginning to emerge across the industry—where AI participates in improving itself and, if left unconstrained, could exceed human capacity to understand and control these systems.

This also explains why he included "embedded independent evaluation" in his proposal rather than merely issuing a safety pledge. Anthropic stated that Accenture's embedded evaluators will closely track how AI models evolve during training and observe the internal decision-making processes behind Anthropic's technology development that are not publicly disclosed. The company described it as: "From this position, embedded evaluators can assess how a company operates, verify whether it is fulfilling its safety commitments, and identify blind spots. They can also report incidents, giving the public a fuller, more informed understanding of both benefits and risks."

The two companies expect to invest at least $1 billion in building these capabilities. Anthropic and Accenture PLC (NYSE: ACN) have collaborated previously: last year they formed a multi-year partnership, building a 30,000-person team trained on Claude to help clients deploy AI systems at scale. In March of this year, Accenture also launched Cyber.AI, a security solution powered by Anthropic's Claude models.

Controversy: How Independent Is This "Third Party"?

The "third-party" label is likely to attract skepticism—Accenture's close ties with major AI startups make it difficult to view as a wholly detached outsider. Earlier this year, the consulting giant projected that its business related to key AI partners including Anthropic, OpenAI, and Palantir would more than double over the next year. In other words, the party brought in to evaluate is also a beneficiary of the relationship, a point that is likely to face repeated questioning once the story is published.

The original report did not specify how evaluation findings would be disclosed, nor did it provide details on the size of the evaluation team or the duration of their work.

Why Now: Resignations, Calls for Slowing Down, and "AI Building AI"

Anxiety over AI existential risks intensified last week following a high-profile resignation. Anthropic employee Jacob Coxon posted a resignation notice on social media, accusing AI companies of "gambling with our lives." According to earlier public reports, Coxon previously worked on pre-training research at both OpenAI and Anthropic and published his resignation statement on X on the evening of September 8, subsequently covered by multiple media outlets.

In the days that followed, leaders from several AI companies, including Amodei and OpenAI CEO Sam Altman, urged a slowdown in the pace of the technology's advancement to address its increasingly unpredictable risks—though they diverge on how to handle it specifically. Amodei has suggested that companies should open themselves to embedded independent evaluators to oversee the development process, and the Accenture partnership serves as the concrete embodiment of that proposal.

Signs of division within the industry have been visible previously: in July, more than 1,100 employees from leading AI companies including Anthropic and OpenAI signed a petition urging the U.S. government to support a mechanism for "deliberately pacing" AI development. A report Anthropic released Thursday (September 10) offered another angle: 26% of the company's AI research and development work is driven by its Claude chatbot. That figure itself also functions as a narrative for the IPO—it points to efficiency while simultaneously underscoring the underlying concern that "AI is accelerating the creation of more powerful AI."