Crowds congregate outside of NASDAQ headquarters in Times Square in New York on 12 June 2026 for the debut of the SpaceX IPO. The IPO price was set at $135 a share, making it the largest IPO in history and making Elon Musk the world’s first trillionaire. reutersconnect.com/Richard B Levine

While IPOs for AI companies are set to attract trillions of dollars, they will also change the technology’s trajectory, with heightened corporate accountability and increasing scrutiny of societal costs and supposed benefits.

When rocket, satellite and AI company SpaceX made its debut on the stock market in June, it became the largest IPO in history, pushing the company’s valuation to over $2tn, and making Elon Musk the world’s first trillionaire.

The move kicked off what could become a pivotal period, with rivals Anthropic, the parent company of the AI chatbot Claude, and OpenAI, the company behind ChatGPT, having both confidentially filed draft IPO documents with the US Securities and Exchange Commission. Neither company has set an official date for a listing, but Anthropic is widely expected to launch on the Nasdaq or NYSE as early as October 2026, with OpenAI following soon after. Both are expected to debut at over $1tn.

The funds generated are necessary to build bigger AI models, creating foundational platforms, similar to Windows for PC or Amazon Web Services.

IPOs are also a major test of investor confidence in a sector that has faced numerous setbacks and intensifying public scrutiny, leaving companies highly exposed. Anthropic and OpenAI are both grappling with multiple copyright, safety and governance lawsuits. For example, in September Anthropic agreed to pay an unprecedented $1.5bn to settle a class action lawsuit filed by authors who said the company stole their work to train its AI models. Experts suggest that, had Anthropic not settled, a potential loss in court could have crippled or ended the company.

As a Public Benefit Corporation, Anthropic has worked to embed ethical governance into its structure – with trustees to safeguard public interest, as well as the implementation of measures which address what might be described as governance issues.

In June, Florida became the first US state to sue OpenAI over the design and safety of ChatGPT. The civil lawsuit, brought by Attorney General James Uthmeier, alleges that the company knowingly released and aggressively marketed ChatGPT to the public, including to children, while concealing serious risks, suppressing internal safety warnings and deceiving Floridians about the true nature and dangers of the product.

In a statement responding to the suit, OpenAI said it has ‘put in place industry leading protections and policies’.

Another potentially destabilising problem is the fact that OpenAI and Anthropic are both operating at significant losses due to intense capital expenditure on AI development, including enormous data centres, computational power and an escalating competition for talent. Anthropic has some of the largest contracts in the sector, for example committing to pay SpaceX $1.25bn per month for access to its computing infrastructure. OpenAI is burning through billions annually on research and development, and profitability is not expected until at least 2029 or 2030.

Current AI infrastructure spending is comparable with the internet build-out of the late 1990s and early 2000s, when early heavy capital outlay was required to secure first-mover advantage. But there are questions when it comes to AI given that, for example, Chinese open-source AI models promise similar performance for a fraction of the price. Much can change swiftly in tech.

The transition from private firms to publicly listed entities comes with significant strings attached, fundamentally shifting AI from a loosely regulated, cash-burning venture phase into a highly scrutinised, revenue-driven public model. Research by Deutsche Bank found that this will change the AI boom forever, bringing simultaneously unprecedented funding and accountability.

Increasing transparency

When preparing to list, AI companies must produce public-company-grade reporting, credible performance indicators, risk analyses and forecasts. Lawsuits over copyright, alleged safety negligence and corporate governance will have to be disclosed. Remuneration for management and employees may need to be revamped to comply with public metrics.

Deutsche Bank argues that greater transparency will ‘finally illuminate the business models of the frontier model makers’ given that monetisation is currently ‘opaque’ and ‘private model makers make only limited disclosures’. It should also give investors a more data-driven understanding of the return on investment.

The legal landscape around IPOs encompasses a range of obligations related to securities laws, regulatory filings, financial reporting, corporate governance, insider trading, antitrust and competition laws and investor protection.

Checks and balances are clearly critical given the trillions of dollars of capital at play. Gabriel Zucman, a professor of economics at the University of California, Berkeley, has warned of the profound effects the consolidation of capital brought on by the SpaceX, OpenAI and Anthropic IPOs, not just for the economy but for society as a whole. ‘There is a fundamental tension in democratic societies between extreme wealth […] and the very possibility of a well-functioning democracy,’ he says. ‘After World War II, it looked like extreme wealth belonged to the past,’ but now, ‘the AI boom is minting billionaires by the day’ and the first trillionaires are coming into view.

The issue raises important questions about who should ultimately own AI, who benefits from it and whose interests it serves. Instead of enriching early private investors via traditional IPOs, alternative models designed to ensure that citizens share the wealth that AI technologies generate are gaining traction.

In the US, there’s a debate about the federal government taking an ownership stake in leading AI companies. According to a Financial Times report, OpenAI’s CEO Sam Altman is in talks with US President Donald Trump about giving the US government a five per cent stake in the company.

Taxing big tech

Senator Bernie Sanders has gone a step further, proposing the American AI Sovereign Wealth Fund Act, which would impose a one-time 50 per cent stock tax on large AI companies. The tax would create a $7tn public fund, managed by an independent commission, with five per cent of the fund’s value distributed annually to Americans.

Justification for the tax is based around the assertion, by Sanders and allied economists, that AI models are built entirely on the knowledge, creative writing, artistic labour and data produced by humanity over centuries. They claim that because tech companies ‘scraped’ this data without explicit public permission or financial compensation, they should not be allowed to retain all of the equity.

Others argue that a 50 per cent tax could stifle innovation and discourage technology investment. Those concerned about environmental impact, including water consumption and the climate crisis, or the misallocation of funds, suggest this would be no bad thing.

Consumer and investor concerns over the societal dangers of increasingly powerful AI have prompted some tech firms to look beyond mere profitability as a driver for business and should look to embed social responsibility and human-centric values into organisational structures.

Microsoft, Anthropic, IBM and Google are among those leading the sector by establishing formal governance models and internal policies to enforce ethical AI frameworks and clear disclosure practices.

According to the Chartered Governance Institute UK & Ireland (CGIUKI), Anthropic has taken arguably the most formalised, internal approach, having worked to embed ethical governance into its corporate structure, training process and decision-making from day one. The company is structured as a Public Benefit Corporation, which is legally committed to long-term human-centred goals. A Long-Term Benefit Trust places power over the company’s future direction with a group of trustees charged with representing the public interest. The Trust can appoint or dismiss corporate board members based on adherence to safety.

Google’s DeepMind operates under a set of published AI principles, which guide the responsible development and application of AI across Alphabet, emphasising social benefit, safety, fairness, privacy and accountability. Formal governance includes internal ethics reviews, fairness audits and the development of open-source governance tools.

Although Google’s ethical governance is structured and serious, the CGIUKI points out that it is ultimately controlled by the commercial priorities. Meta has arguably been more commercially driven and less transparent than other major AI players. Although it has established AI ethics teams and published commitments on fairness and bias, the CGIUKI said ‘its strategic direction appears to prioritise scale, speed and competitive positioning’ and ‘it has not released detailed frameworks for AI alignment or risk governance’.

The AI boom is minting extraordinary fortunes. But, creating extraordinary public value through policies that redistribute the wealth generated, and commitments to deploy AI in the public interest, remain much bigger and more important challenges.

Stephen Cousins is a journalist covering the built environment, sustainability and technological innovation.