Anthropic, the company behind the artificial intelligence platform Claude, is preparing to go live with its initial public offering on Thursday, but researchers and investors are raising red flags about the company’s potential ESG risks.
Anthropic initially filed for an IPO in a confidential securities filing in June, leading investors to speculate about the company’s plans to address existing governance and transparency concerns. The company’s S-1 filing — which registers securities for public sale with the Securities and Exchange Commission — was released late last month, giving investors a little under a month to dig in and find any potential risks to their investments.
“The key thing that we've learned is that there are a lot of risks that the company is going to have to keep an eye on, both from within the business, but also their interaction with the rest of society, government [and] a lot of other stakeholders,” Morningstar Director of Institutional Insights Lindsey Stewart told ESG Dive in an interview. “And I guess those are the key questions that institutional shareholders have to ask.”
The AI firm’s public offering plan includes a unique governance structure that Stewart and investors ESG Dive spoke to believe could insulate the company’s founders from shareholder pressure. Additionally, the company lists a litany of risks the company’s operations face; the prospectus lists over 90 pages of potential threats to the company’s operations.
On Monday, Morningstar’s ESG research arm labeled the pending stock release a “high ESG risk,” while investors and pension funds have raised additional concerns about the company — from its governance structure and potential risks to what Anthropic’s disclosed financials reveal.
What the ‘high risk’ rating means for investors
Ahead of Anthropic’s IPO launch this week, Morningstar Sustainalytics labeled the stock “high risk” in an ESG Risk Rating Report shared with ESG Dive. The rating comes as Morningstar scored the company poorly on “material ESG issues” of business ethic and human capital, according to the Nov. 16 report.
Additionally, the report found that the company has a medium exposure to material ESG issues — a level of exposure Morningstar said is “is significantly above ... average” fellow enterprise and infrastructure software providers — and exhibits average management of those issues. The report flags “severe” ESG risk for the company for business ethics and a “high” risk for the company in terms of human capital. The Monday report also flags high exposure for the company in areas of corporate governance, business ethics, human capital, product governance and data privacy and cybersecurity.
“Ensuring the safety, reliability and responsible use of AI presents significant challenges for AI providers such as Anthropic,” the report said. “As Anthropic scales, its innovative governance structure combining the Long-Term Benefit Trust and founder-concentrated control will have to prove effective at balancing safety and commercial pressures.”
Morningstar said Anthropic has a high exposure to ethical risks “due to its reliance on large-scale data collection and web scraping.” Morningstar said the company’s management of such risks is “weak,” as it does not have programs in place to disclose business ethics risks like bribery, corruption or conflicts or interest and it’s “unclear what mechanisms Anthropic has for stakeholders to report illegal or unethical activity.”
“Anthropic’s dependence on infrastructure and compute providers that include investors, suppliers and competitors further amplifies risks of potential conflicts of interest and governance challenges,” the report states.
The company scored “strong” on management of its product governance risk, another area of high exposure for the company. While the company has had “numerous controversies on third-party misuse and uncontrolled model behavior,” Morningstar noted that Anthropic has a board-level oversight committee for such issues and consistently reports incidents.
Anthropic’s stock structure sparks governance concerns
After first disclosing its S-1 filing on Oct. 26, Anthropic has since amended its filings twice on Nov. 2 and Nov. 10. The company is preparing to be domiciled in the state of Delaware with a governance structure that includes five different equity classes.
While Anthropic is issuing 801,351,354 Class A shares, other existing stockholders are selling another 23,648,646 shares — bringing the total offering to 825 million shares priced between $70-78 each, according to its Nov. 10 amended filing.
While those Class A shares will come with a traditional vote equity, the company’s seven co-founders collectively control 50.1% of “whatever the voting power is” on general shareholder matters through a single voted share that can’t be diluted, according to Stewart.
The governance structure also includes a stock class for director elections and an equity class to change the company’s Long-Term Benefit Trust.
“The novel share structure that Anthropic has put through has been an important topic of conversation,” Stewart said. “It does appear that the technology sector, in particular, is not receptive to the idea that equal voting rights is the way forward in the way that institutional investors insist on.”
Stewart’s comments were echoed by SOC Investment Group’s Research Director Richard Clayton, who told ESG Dive in an interview this month that Anthropic’s governance rules are “ set up in a way which really insulates the company's key decision makers from any kind of accountability to the company's shareholders.”
SOC Investment Group is the shareholder engagement arm of union federation the Strategic Organizing Center, and Clayton said that Anthropic’s justification is that the structure will allow the company’s decision-makers to make choices for what’s best for society rather than profit.
“The problem is that right now we've been through several months of increasing concern around the agentic AI models not being really under the control of the companies that have created them,” Clayton said. “But rather than slowing down, Anthropic's moving ahead with the IPO.”
Investors flag broader financial, ESG issues
Beyond a poor ESG risk rating from Morningstar, investors ranging from institutional, to sustainability-focused to public pension funds and groups representing unions have also raised issues with the pending go-public plan.
“Anthropic is going public at a moment when communities everywhere are grappling with the consequences of AI, from data centers in rural towns to the future of many jobs, while headlines warn of existential risk,” Josh Zinner, CEO of the Interfaith Center on Corporate Responsibility, said in a media briefing last week.
Anthropic itself raised four major categories of risks the company faces in its S-1 filing: “misuse, loss of control, authoritarian abuse, and economic disruption.” The company’s prospectus also includes over 90 pages of risk factors for the firm, including its reliance on computer power and data center investments, as well as electrical power and fuel constraints, including in the U.S.
“Our success depends in large part on our ability to continue to develop the most advanced and capable AI models, and any failure to do so could adversely affect our business,” Anthropic said in its IPO registration.
SOC Investment Group sent a letter to potential investors last week, warning that the company’s expected valuation has fallen since its registration statement went public, dropping from early estimates near $2 trillion to around $1.4 trillion as of Nov. 10.
Anthropic posted an operating profit of $974 million for Q2 of this year, but the profitability came while the company was receiving a discount on its data center agreement with Space X, according to Clayton. Anthropic’s amended Nov. 10 registration filing shows that the company is operating at a $567 million loss, with net loss of $104 billion, for the nine months that ended Sept. 30.
Clayton said the company’s financials revealed that “a substantial part” of its revenue comes from customers with an annual revenue of less than $10 million. Because of that, he said it is unclear if Anthropic can keep up its current spending on its models.
“There are a lot of partners that are financially interested in the business in one way or another,” Stewart said. “It looks like there is a lot wrapped up in the success of this IPO for a very very broad range of companies, not to mention the VC world and a lot of other stakeholders wrapped up in this, so it's really a big moment for the entire AI movement.”