GIXI

3.0201.31%

Anthropic IPO signals AI market maturity, but investment bubble poses risks: OpenAI chairman

LongbridgeAII'm LongbridgeAI, I can summarize articles.

[SINGAPORE] Anthropic’s impending initial public offering signals a maturing artificial intelligence market, but a surge in speculative investment across the industry is also creating risks, said Bret Taylor, chairman of OpenAI on Wednesday (Nov 18).

Taylor, who is also the co-founder and CEO of enterprise AI company Sierra, welcomed Anthropic’s move to go public, saying it would allow more investors to participate in the growth of AI.

“I think it’s good that Anthropic is going public,” he told The Business Times in an interview on the sidelines of the Singapore Fintech Festival.

Anthropic, the creator of AI large language model Claude, is set to make its Nasdaq debut on Thursday, in what could be one of the largest technology listings in history. The company is targeting a valuation of about US$1.4 trillion.

OpenAI has also signalled its intention for a potential listing in the coming year, with Taylor noting that going public is “something OpenAI will obviously consider in the future”.

He said that the broadening ownership of AI technology “demonstrates the sort of maturity of the market overall”. “So I’m really happy for the folks at Anthropic for reaching that milestone.”

At the same time, Taylor acknowledged that the industry is experiencing an investment bubble, even as he remains positive that AI will generate significant economic value.

“We’re in a bubble – that is, there’s not just smart money but dumb money, and we’re delivering real economic value,” he said. The opportunity and challenge for investors, he added, would be to discern between sound investments and those taking excessive risks.

Bubble fears

Taylor noted that the rapid expansion of AI infrastructure investment has created conditions resembling the dotcom bubble, when many companies collapsed even as the Internet boom gave rise to technology giants such as Amazon.

AI could follow a similar path, with some companies failing despite the technology’s broader potential, he added.

He drew parallels to what is happening today with the AI boom, where some investors and companies “take a lot of risk and take a lot of leverage”.

Among the more vulnerable businesses are companies that serve other AI companies exclusively, and with individual data centres funded with debt, he said.

Still, Taylor said that there is a “shortage of compute”, exacerbated by vendors who have a dominant position in the market.

“There’s not enough data centres. There’s not enough power,” he said during a fireside chat at the Singapore Fintech Festival. “Almost every part of the stack of AI is a scarce supply.”

He urged countries to treat data centres and compute production as critical resources that could ultimately drive economic growth. “Just like energy, we never really have enough,” he said. “I just hope we build more and more and more.”

Profitability and other AI risks

As the potential listings of AI giants loom, market observers have raised questions about these companies’ path to profitability.

“It wouldn’t be appropriate, obviously, to talk about our financial numbers,” said Taylor, who declined to give a timeframe for OpenAI turning a profit.

However, he split the economics of the model maker into two parts: the margins it earns on selling a model once it is trained, and the fixed cost of training it. “I think there’s a real credible path there to develop a really healthy business,” he said.

Taylor reiterated that OpenAI is “not just in it for the profits”, noting that the firm’s mission is to ensure that AI “benefits humanity”.

Apart from financial risks, Taylor believes that the next 12 to 18 months could be a period of heightened cybersecurity vulnerability as malicious actors potentially use AI to exploit weaknesses in existing systems.

He noted that AI makes it easier to identify and exploit vulnerabilities, with companies often unable to keep pace with the need to patch their systems.

“The uncomfortable truth is all of these vulnerabilities existed before AI, and AI just shone a light on them,” he pointed out. “It’s easier to create chaos than it is to patch systems.”

He emphasised that companies must remain accountable for the actions of AI agents deployed on their behalf. “I think AI is a tool, and so if a company uses AI, I think they’re accountable, just like if an employee did it,” he said. “I don’t think using AI changes accountability.”

Despite the risks, Taylor expects the industry to be in a better position 18 to 24 months down the road, as AI-driven defensive capabilities develop that will help organisations strengthen cybersecurity.

Addressing the chip exports

When asked to weigh in on the latest US chip export controls, Taylor said that they were unlikely to fundamentally complicate Sierra’s business in Singapore and the wider region.

Sierra is an enterprise conversational AI company that uses AI agents to handle customer service. “I’m sure it does in nuanced ways, but at the high level, no,” he said.

Under the new rules announced on Monday, exporters will need a licence to ship advanced AI chips to Singapore, Malaysia, Indonesia, Thailand and Brazil.

Nevertheless, Taylor cautioned that the measures are still new and their full implications remain uncertain.

When asked if the curbs could leave affected countries at a disadvantage in AI, he said: “I don’t think so.” He argued that while computing capacity remains one of the biggest constraints on AI, the bottlenecks go beyond chips to power and data centre infrastructure.

He believes these constraints could spur innovation, including new AI model architectures that require fewer computing resources and alternative approaches to chip manufacturing.

“I do think constraint breeds innovation,” he said.

For Singapore – which is short on land and power – Taylor suggested approaching these challenges “from first principles”, beginning with addressing the power needs.

He suggested fusion as a potential solution, noting that it “will be a while, but maybe shorter than we think”. Still, he cautioned that such technological advances would not be straightforward.

“I don’t think it happens overnight, nor do I think it’s easy,” he said.

Source: Businesstimes News The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.