Anthropic Mulls New AI Model Amid Investors’ Pre-IPO Worries

Anthropic is reportedly considering the launch of a new artificial intelligence model before it goes public.

That’s according to a Friday (Sept. 18) report by Reuters, citing three sources familiar with the matter. The report noted that a new AI model could offset the momentum OpenAI has enjoyed since debuting its GPT-Astra.

Released earlier this month, GPT-6 Astra has gained momentum among business customers, leading investors to rethink Anthropic’s position as the leader in making enterprise AI tools, the report said. One source told Reuters Anthropic is evaluating the next model’s safety as it deliberates a possible release.

Some of those deliberations center on balancing investment in deploying new models with bolstering the company’s profitability, as higher interest rates make backers more focused on the timing of anticipated profits, the report said, citing sources familiar with Anthropic’s thinking.

A separate report by the Financial Times said investors are wondering whether Anthropic can maintain its growth in the face of competition, customer price sensitives and concerns about AI safety.

The startup’s annualized revenue reached $65 billion in July, and its backers predict it will top $120 billion by the end of the year, making Anthropic the fastest-growing business of all time. However, the report said, there is still considerable uncertainty over whether the company can sustain that pace, as seen from the wide range of IPO valuations estimated by investors familiar with the group’s performance: between $1.5 trillion and $4 trillion.

PYMNTS has contacted Anthropic for comment but has not yet gotten a reply.

Anthropic had been targeting late October for its initial public offering (IPO), but is now shooting for November, according to a Friday Wall Street Journal report.

The news also follows Anthropic CEO Dario Amodei’s recent call for the AI industry to scale back the launch of new capabilities amid safety concerns.

“We must slow the pace at which we improve the capabilities of AI models,” Amodei wrote in a 3,800-word essay Sept. 12 that paints a dire picture of a world beset by AI-related attacks.

“Given the accelerating rate of AI capability development, it’s my worry that in 6–12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails,” Amodei wrote.

PYMNTS spoke last week with experts about Amodei’s call for a slowdown. Among them was Maya Mikhailov, CEO and co-founder of Savvi AI, who views a gap between what the labs are worried about and what AI companies’ customers actually use.

Slowing frontier AI development is not akin to slowing AI adoption, she told PYMNTS, because most enterprise customers “aren’t even using the state-of-the-art models to begin with.”

That gap can be seen in the revenue as well: Ramp’s AI Index found the top 1% of businesses are behind the bulk of model companies’ enterprise revenue. It also gives the labs a financial incentive to ease back.

Training these models costs billions of dollars, said Mikhailov, “while most of their enterprise clients aren’t even coming close to using the advanced models’ capacity.”

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.