Welcome to another tech earnings week in the United States where Apple, Meta, Microsoft and Tesla prepare to issue market-moving financial results. Could there soon be another mega tech name added to that Magnificent Seven list whose weight accounts for a third of the S&P 500?

The rumour mill has been working overtime, speculating that OpenAI or Anthropic will float this year, giving the starry-eyed public a chance to own a piece of the hottest companies on the planet. The reality is less straightforward.

It is true, these great beasts of generative AI require huge amounts of capital to fund their quest for the holy grail of the most advanced forms of the technology. This is not a bootstrap endeavour.

OpenAI needs well over a trillion dollars to cover its pledges for investments in data centres and AI chips, its aggressive expansion into everything from healthcare and ecommerce.

Yet launching on to the AI-thirsty equity markets is not yet necessary for the makers of ChatGPT and Claude, because the cash continues to gush into them with the intensity of a firehose.

OpenAI was valued at $500 billion in October after an employee share sale saw investment by Softbank and MGX of Abu Dhabi. Just last month the chatter was that OpenAI was in another round of talks to raise another $100 billion which would value it at $750 billion. Similarly, Anthropic is on the verge of being given a $350 billion price tag as it happily raises $10 billion or more.

These businesses still have far too many unanswered questions to go public. Staying private keeps a comfortable cloak of secrecy over their operations and gives them time to unpick their devilishly complex corporate structures.

Being a listed business would require an unprecedented level of transparency, not least over whether their revenues can catch up with their spending.

Sure, they are growing at a rate of knots, but we are not quite sure at what cost. Anthropic’s revenues doubled from $4 billion to $9 billion last year. OpenAI’s annualised revenues (what the company is earning right now, scaled up as if that pace continued for a full year) rose above $20 billion in 2025, up from $6 billion in 2024.

Yet the mammoth outgoings on computing capacity dwarfs sales, and they need to keep spending because bigger models create more user demand and support more paying customers.

While public markets might raise eyebrows, none of this uncertainty is putting off the private markets. Asked by Brad Gerstner, his own investor, how OpenAI could commit to spending trillions while earning merely billions, Sam Altman, OpenAI’s chief executive, shut him down with sharp confidence.

“If you want to sell your shares, I’ll find you a buyer,” Altman snapped on the BG2 Pod. Who needs retail shareholders when private buyers are baying for equity?

• Will AI bubble trouble the stock markets in 2026?

A listing would mean the grind of quarterly reporting, exposing their detailed financials or as OpenAI’s chief financial officer, Sarah Friar, called it, getting “wrapped around an IPO axle”.

Unlike their private British counterparts who make annual filings with Companies House, the Americans need to reveal very little of what goes on beneath the bonnet to anyone, except their leading investors.

Perhaps more crucially, a float would mean revealing the details of the complex circular deals which have raised the spectre of the dotcom boom and bust. It would provide an answer to that well-worn question “are we in a bubble?” and it may not be an answer they want to give.

There’s a nervousness about being the first mover and suffering a spectacular flop. Then there’s the force majeure. The US federal government shutdown last year hurt IPOs as the administrative machine stalled while President Trump’s erratic behaviour is churning up the markets.

In the meantime, anyone keen to get exposure to these shares can do it the long way round, by buying into their heavyweight partners, Amazon or Microsoft. At least until OpenAI or Anthropic decide they want public market money badly enough to tolerate public market questions. Right now, that feels some way off.