The Story
Technology giants, chip companies and AI cloud providers have sold or priced more than $300 billion of new stock and mandatory convertible securities during 2026, according to The Information, with roughly $100 billion of further share sales still available or in registration.
The tally excludes Anthropic's planned initial public offering, reported as a raise of up to $100 billion, which would take total equity supply from the sector past $400 billion. Mandatory convertible securities, which must convert into equity at maturity rather than leaving the choice with the holder, are included in the figure.
The report follows a volatile week for AI-related stocks, which has drawn attention to investor appetite for additional offerings and to the prices companies would achieve if they chose to sell.
The equity issuance runs alongside a parallel expansion in borrowing. Microsoft priced $25 billion of senior unsecured notes on 29 October, its first US investment-grade bond sale since 2017, drawing more than $115 billion of orders. Alphabet raised $25 billion in August and Amazon $25 billion in July. Estimates of total AI-related debt issuance for 2026, extending beyond the hyperscalers to chipmakers, data centre developers and utilities, range from roughly $300 billion to $570 billion.
By comparison, US initial public offerings collectively raised about $267 billion across the five years to 2025.
Why It Matters
Set this number against the one the Bank of England published four days ago and the gap is the entire argument.
British firms told the central bank that three years of using AI had raised their productivity by about 0.5 per cent, roughly 0.17 percentage points a year. In the same period, and principally in the last twelve months of it, the companies building AI have raised more than $300 billion of equity, have another $100 billion registered, and sit alongside an estimated $300 billion to $570 billion of sector debt issuance. Somewhere between $700 billion and $1 trillion of external capital, in one year, against a measured return that a central bank describes as modest.
That is not automatically an indictment. Railways, electrification and fibre all absorbed enormous capital long before the productivity statistics noticed, and each eventually justified it. Infrastructure is supposed to be front-loaded, and the Bank itself said the contribution so far reflects the building rather than the using, which is exactly what an early buildout looks like.
The difference is who is being asked to carry the gap. In previous cycles the lag between spending and return sat on the balance sheets of a small number of operators and their lenders. This time it is being distributed, through public equity to index funds and pensions, through investment-grade bonds to insurers, and through mandatory convertibles to anyone holding the paper when it converts.
Which makes the absorption question the real one. Not whether AI eventually produces the returns, but whether the market keeps buying the supply during the years before it does, and at what price it stops.
The Strategic Read
The instrument mix is the part that rewards attention.
Mandatory convertibles sit inside this tally, and they are not an ordinary choice. A company issues one when it wants cash now but does not want to print shares at today's price, and when it values the partial equity credit that rating agencies extend to paper that is contractually obliged to become stock. They are the instrument of an issuer who needs equity, is not happy with the price, and has decided to take the money anyway while deferring the dilution.
Seeing them in size tells you something that the headline number does not. The straight versions of both financing routes have become more expensive at the margin, so issuers are reaching for the hybrid in between. That is a late-cycle pattern rather than an early one.
The absorption question is then arithmetic rather than sentiment. Roughly $400 billion of equity supply, counting what is registered and what Anthropic is reported to be seeking, has to find buyers in a market where the same institutions are also being asked to take several hundred billion dollars of AI-related bonds. Every one of those offerings competes for the same finite pool of institutional attention, and the ones that price last price worst.
For Indian issuers the read-across is about queue position rather than quality. Several large listings are being prepared here for 2027, and they will be marketed to global allocators whose AI exposure has just been topped up substantially, at prices set before last week.
For daily, sharp analysis of the biggest moves in the Indian business and startup ecosystem, follow StartupFox.