Anthropic today became the first AI model maker to go public, just a week shy of the fourth anniversary of the launch of its archrival OpenAI’s pioneering ChatGPT, bolstering the value of AI for enterprises that are still struggling with the ROI of the technology and overshadowing AI competitors.
The historic public offering from the creator of the Claude line of generative AI models started at $110.50, 34% above the IPO price set on Wednesday. It raised about $81.3 billion across 980 million shares, shattering SpaceX's previous $75 billion record.
With the vendor's first day trading on the public market - under the stock ticker symbol ANT - exceeding expectations, it also provided some assurance across the AI industry that financial value lies in the technology, even as enterprises are still scurrying to demonstrate ROI in how they have incorporated it.
"The fact they got way above the offering price should give enterprises confidence that enterprise AI is a real thing," said Nick Patience, an analyst at Futurum Group. "For CIOs in some ways it lowers the risk of betting on the company."
The dramatic public offering comes after Anthropic initially predicted its valuation would be about $1.33 trillion. The actual value ended up at $1.5 trillion.
That initial devaluation in early November followed the Nov. 3 mid-term elections, which revealed widespread negative sentiment in the U.S. about AI and AI data centers. It also comes amid an AI market grappling with the reality of expensive technology that has mostly yet to yield a return. The lack of immediate ROI has led AI vendors, including Anthropic, to try to pacify enterprise concerns by cutting prices.
Not only did Anthropic’s first day being listed on the Nasdaq stock exchange give some measure of certainty to enterprises, but it also provided even more visibility to vendor’s technology.
What the new funding means
The generative AI startup has spent the six years since it was founded focusing on the enterprise market, compared to OpenAI, which has only lately made serious inroads with businesses.
So, the record-breaking IPO suggests that Anthropic’s approach is correct. Its successful debut as a public company means it can now continue to purchase compute and raise debt easily.
"Anthropic was already sitting on more than $100 billion before the IPO," Patience noted. "The new money buys it use over its supply chains."
He added that Anthropic can now step into some of the deals that before appeared uncertain, using prepayments or guarantees to secure capacity.
"Over time, it could own more of its infrastructure outright," Patience continued.
He was referring to the fact that currently Anthropic mostly holds contracts in which it rents compute and cloud services from vendors such as AWS, Google Cloud and Elon Musk’s SpaceXSI. Now, with access to more capital, the vendor could one day buy instead of lease the compute it needs.
However, going public also means Anthropic needs to slow down its participation in the price war, he said. Along with OpenAI and others, the vendor has been lowering the prices of its models to compete with models from open-weight vendors - most of them from China - whose models are on par with some frontier models.
"They will continue to cut those prices, Patience said. “I wonder if they will cut them maybe at a slightly slower rate than they were going to because Wall Street will put a new kind of pressure on the company that they've never had before,"
Other than using its new $81.3 billion cash storehouse to buy compute power, the AI vendor will also be encouraged to create more advanced models and other new technology and products that build on AI, said Michael Bennett, associate vice chancellor for data science and AI strategy.
"It will likely be incentivized to do so to increase shareholder value," Bennett said.
Enterprises should be careful
However, Bennett cautioned against viewing Anthropic's day-one success as a definitive indication of "the market's appetite for a pure AI play."
"Right now, it seems premature," he said.
He suggested that enterprises respond soberly, remain clear about what technology means for their businesses, and use it accordingly.
A clearheaded response by enterprises to the blockbuster IPO makes sense because of the unpredictability of AI technology, said Brian Oh, global head of technology media and telecom at Guidepoint, a network and information services company.
"If you look at a 10-year [outlook], this is a lifetime opportunity. But then for the short term, there's too much volatility,” Oh said. “It's hard to guess which direction it's going to head,"
Focus on cybersecurity
Along with not getting carried away with the excitement of the IPO, companies should reinforce their commitment to cybersecurity, Bennett said.
“Enterprises should continue maintaining a vigilant stance toward the development or the release of the models," he continued.
He said enterprises should closely monitor frontier AI vendors and consult cybersecurity experts when a big AI vendor such as Anthropic releases a powerful new model, and when reports indicate that safety tests reveal weak spots.
"We're probably locked into an arms race between advancing models that can find weaknesses in existing cybersecurity arrangements and digital systems generally and the industry's ability to remedy those defects," Bennett said.
Anthropic's IPO could also be a preview of what's to come for rivals such as OpenAI, which is preparing to go public next year.
Esther Shittu is a TechTarget reporter and podcast host and producer covering AI systems.