The AI buildout has an appetite for capital that keeps surprising even the people funding it. Every quarter brings a bigger number, wider spending guidance, and a fresh round of questions about where all that money is actually coming from.
One of Wall Street‘s largest banks just tried to answer that question directly. And the numbers behind the answer say more about the health of this rally than any single earnings report has this year.
Goldman Sachs on record $252 billion AI equity issuance surge
U.S. companies raised a record $252 billion through IPOs, follow-on offerings, convertible securities and SPACs in the second quarter. That topped the previous all-time quarterly high of $234 billion set in Q1 2021, according to Seeking Alpha, citing Goldman Sachs strategist Ben Snider.
AI-related companies made up roughly 40% of follow-on equity volume so far this year. Year-to-date follow-on activity through July hit its highest level for the same period since 2021, reaching $105 billion.
That surge traces directly back to infrastructure spending. Goldman Sachs estimates the four largest hyperscalers will spend roughly $725 billion on capital expenditures in 2026. That is roughly equivalent to nearly all of their combined cash flows from operations. Capital spending is expected to outpace operating cash flow by a wide margin in 2027.
More AI:
- Nvidia just made a move Wall Street wasn’t ready for
- Microsoft just took sides in AI policy fight
- OpenAI just disclosed something genuinely alarming
Debt, not new stock, is expected to carry most of that funding gap. Goldman Sachs credit strategists project hyperscalers will finance more than a third of their 2027 capital expenditures with debt. That translates into hundreds of billions in global bond issuance next year alone, according to Yahoo Finance.
Equity still plays a role in that financing mix, even if it is not the primary source. Selling shares gives companies additional funding without putting as much pressure on their balance sheets. Issuance tends to pick up when the market is performing well and stocks command premium valuations.
Both of those conditions have held through this stretch of the AI buildout.
Why Alphabet’s free cash flow turned negative for the first time
The clearest sign of strain showed up at Alphabet.
The company raised its 2026 capital expenditure guidance to $195-$205 billion, up from a prior range of $180-$190 billion. It reported negative free cash flow of $5.9 billion for the first time since going public in 2004. Capital spending of $44.9 billion in the quarter outran operating cash generation of $39.1 billion, as TheStreet reported.
Alphabet also repurchased no stock in the quarter, against $13.2 billion in buybacks a year earlier. It redirected that capital entirely toward infrastructure.
Alphabet is not alone. Amazon’s trailing 12-month free cash flow swung sharply negative as property and equipment spending jumped by more than $66 billion. Combined capital expenditures from Microsoft, Amazon, Alphabet and Meta are expected to reach roughly $725 billion this year.
That is up about 77% from $410 billion in 2025, according to Seeking Alpha.
Rochlin/Getty Images
S&P 500 buybacks vs. the AI capex wave draining tech cash flows
Corporate buybacks have historically absorbed new share supply. Goldman expects that pattern to hold for now.
S&P 500 buyback growth was tracking at double-digit rates year over year in the second quarter. Total U.S. buyback authorizations reached a record high through the date of the report. Full-year repurchases are projected to reach $1.4 trillion. That more than covers direct equity issuance.
The cushion is not spread evenly. The biggest AI spenders are also the ones pulling back hardest on buybacks. Goldman found that repurchase activity among the heaviest AI capex names has slowed as infrastructure spending has climbed.
The Magnificent 7 account for a meaningful share of gross S&P 500 buyback activity. Their collective buyback growth came in flat in the second quarter as cash got rerouted toward data centers and chips.
Other sectors are stepping in. Companies benefiting directly from AI infrastructure demand, along with financial firms facing an easier regulatory environment, have been increasing their own repurchase activity. Goldman expects that trend to continue supporting the broader buyback total, even as hyperscaler activity cools.
Goldman Sachs S&P 500 forecast and the $700 billion supply question
Goldman expects 2026 to set a dollar record for U.S. corporate equity issuance, forecasting roughly $700 billion in total supply. Goldman partner John Flood notes that figure represents only about 1% of the Russell 3000’s market capitalization. That is in line with the historical average from 2015 through 2019.
Flood says the S&P 500, which recently closed at a record 7,757, could still climb toward 8,000.
Investors appear to be absorbing the current wave of supply without unusual difficulty. Follow-on offerings have priced at an average discount close to historical norms. The median stock affected by a new offering has typically recovered within a few months.
The picture Goldman is describing is one where AI spending keeps pulling in outside capital at a record pace, without yet creating an equity glut that could overwhelm demand. The bank calls it a headwind but not a gale.
For investors trying to gauge how much longer the AI trade can keep running on other people’s money, that distinction may end up mattering more than any single earnings beat.
Related: Microsoft makes a controversial decision that changes its AI story