Anthropic IPO to test the first 'pure play' AI stock in a choppy market

Résumé détaillé

American AI giant Anthropic's expected debut on the Nasdaq has one money manager recalling the dot-com bubble of the 2000s.

Anthropic's stock market debut is set to be big — a big valuation, big spending, big warnings, big uncertainty for investors and a big potential indicator of how the recently volatile artificial intelligence sector will fare in 2027.

In an amended U.S. regulatory filing Tuesday, San Francisco-based Anthropic set an expected initial public offering (IPO) price range of $70 US to $78 US per share. That implies a fully diluted market valuation of roughly $1.3 trillion US at the midpoint, climbing to $1.4 trillion US at the top end, marking the largest technology stock market debut on record.

"It's going to be one of the biggest IPOs in history," said Greg Taylor, portfolio manager at Vancouver-based PenderFund Capital Management.

The size of an IPO is typically measured by the amount of capital raised. Anthropic is aiming to raise up to $64.4 billion US directly through its public share offering. If that happens, it will rank behind only SpaceX's record-setting stock market launch in June.

Taylor expects Anthropic shares will start trading before America celebrates Thanksgiving, on Nov. 26. Wealthsimple, the popular Canadian financial trading app, lists an expected "IPO day" of Nov. 30.

Ernest Wong, head of research at Toronto-based Baskin Wealth Management, anticipates the stock will be a serious heavyweight when it starts trading on the Nasdaq exchange.

"There are only a handful of stocks in the world that are valued over $1 trillion US," he said. "It speaks to how big AI really is."

Anthropic started as a niche AI research lab in 2021, when its founders left rival OpenAI because they felt it was too focused on profits, ignoring risks and giving too much control to investors.

The company gained momentum in 2024, when it released Claude 3.5 Sonnet. This version beat OpenAI's GPT-4o on major performance benchmarks at a fraction of the cost of previous frontier AI models.

Début de l'article, extrait de CBC News. Pour lire la suite, ouvrez l'article chez le média.