enesptjazh-Hanszh-Hantfrdeitkoruarhiidtrnlplvithsvhemsukcselrohudafinobn
Meta's cash flow went negative, so it built a non-voting stock to pay for AI
Meta burned $2.12 billion of free cash in Q3 and created non-voting Class C shares. Tesla lined up $30 billion of credit, and Huang calls capital scarce.
Meta spent $34.91 billion on capital projects in the third quarter and brought in $32.79 billion of operating cash, so free cash flow came out at negative $2.12 billion. On the same October 21 release, Mark Zuckerberg's board created a new class of publicly listed, non-voting stock. Tesla reported the same evening and disclosed $30 billion of fresh credit lines. At Nvidia's Berlin conference that week, Jensen Huang said the neoclouds have plenty of customers and not enough money. For three years the AI buildout ran on cash the big platforms already had. This week the people paying for it started saying out loud that the cash runs out somewhere.

Meta's new stock has no vote and a clear job
The mechanics are simple. Every Class A and Class B holder gets one Class C share per share held, which works like a two-for-one split, and the new charter takes effect January 19, 2027. Nobody loses a vote, because the new shares never had one. CFO Susan Li listed what the stock is for when Goldman's Eric Sheridan asked about it on the call: capital raising, employee pay and acquisitions. She also said Meta has no specific financing plan to announce. I would take that literally and no further. A company builds a share class it can sell without diluting the founder's control because it expects to sell some.
Investors read it the same way. Meta stock fell nearly 7% on October 22, and its five-year credit default swap spread hit an all-time high, even though revenue rose 26% to $64.6 billion and beat estimates. The company also raised the bottom of its 2026 capex range to $135 billion.
Zuckerberg pulled monetization forward
The more telling remark came near the end of the call, when Evercore's Mark Mahaney asked whether Muse would follow Meta's usual order of users first and money later. Zuckerberg said no, and gave cost as the reason.
My guess is that we are going to work on monetization somewhat sooner on this than our normal playbook for a couple of reasons. One is just the size of the numbers that we're talking about in terms of both the costs and the opportunity, I think, make it kind of important to make progress on that in order to be able to scale it.
Mark Zuckerberg, October 21, 2026, Meta Q3 2026 earnings call
For Stories, Reels and WhatsApp, Meta built the audience first and charged later. With Muse, the personal agent that more than 10 million people in the US and Canada have set up, cost is setting the schedule. And the plan to make it pay rests on an efficiency promise Zuckerberg made earlier in the call.
In the near term, Muse is very compute intensive to run, but we expect to improve its cost efficiency by more than 10x in the next year, which will enable it to be profitable once the revenue lines are at scale.
Mark Zuckerberg, October 21, 2026, Meta Q3 2026 earnings call
Ten times in a year is a big number to put on the record. If it slips, the gap between serving costs and revenue has to be filled by somebody, and Meta just built the tool for that.
The analysts heard an equity raise
Gene Munster of Deepwater went on CNBC after the call and did not hedge.
Just before I hopped on, that question got asked, and they said they just want to be ready. Translation is they're probably going to have to raise capital.
Gene Munster, October 22, 2026, CNBC Television
He put a size on it a moment later, "25, 50 billion, something like that." That is his guess, not Meta's. Mahaney was more measured on Prof G Markets but landed in the same place.
And so a reasonable question is, you're doing this to raise capital. How much capital are we talking about? And they're not going to answer that now, but, you know, I assume we're going to get an answer to that very soon.
Mark Mahaney, October 22, 2026, Prof G Markets
Mahaney's other point is the one long-term holders should sit with. He grouped Meta with Tesla and SpaceX as companies where "you're buying into an asset that's owned by the founder," and said at least it is all upfront. A non-voting class makes that permanent for any new money Meta raises.
Tesla chose debt
Tesla went the other way. CFO Vaibhav Taneja cut this year's capex guidance to about $20 billion from more than $25 billion and said the company had lined up borrowing it does not plan to use yet.
Additionally, during the quarter, we secured capacity to borrow up to 30 billion to support the expected growth in capex.
Vaibhav Taneja, October 21, 2026, Tesla Q3 2026 webcast
According to Electrek's read of the 10-Q, that is a $20 billion delayed-draw term loan due 2029 plus a $10 billion revolver, both undrawn at quarter end. The reason for borrowing now is Terafab, the chip plant. When an analyst asked what it would cost, Elon Musk said "tens of billions" through early 2028 and more than $100 billion over time, then added the part that matters while a SpaceX merger is under review.
Although we've got to figure out what the sort of relative split is between SpaceX and Tesla in terms of who pays for what at Terrafab.
Elon Musk, October 21, 2026, Tesla Q3 2026 webcast

The market punished both choices. Dan Ives, now at Yorkville Ives, summed up the bind on BNN Bloomberg.
And I think exactly what you said: it's kind of like either way you go, the market's taking it down, whether it's on the capex light or capex acceleration.
Dan Ives, October 22, 2026, BNN Bloomberg
Huang says capital is now the scarce input
A week ago the shortage was GPUs, then power. At GTC Berlin, Jensen Huang named a new one when he was asked about the neoclouds.
And so what is missing in the middle is the working capital to fuel that growth, to fund that growth. And because there's just so much demand across so many sectors of AI, there's a scarcity of capital.
Jensen Huang, October 21, 2026, GTC Berlin interview
His answer is to make the chips themselves the collateral. "One of the things that we're trying to do is to establish NVIDIA's compute as a financable asset," he said. That pitch only works if lenders believe a GPU holds its value for years, which is the same bet CoreWeave's bondholders have been doubting.
Gray wants more of it public
Jonathan Gray reported Blackstone's quarter on October 22 and was asked whether higher rates threaten the build. His answer was that demand is fine and the problem is the form of the money.
They can do chip financings at the asset level. They may have to raise equity.
Jonathan Gray, October 22, 2026, Bloomberg Television
He went further and called an Anthropic IPO the release valve. His argument is that a listed lab with its own rating can borrow more cheaply instead of leaning on hyperscaler leases. Gray sells private credit for a living, so a man with a product to place is telling you capital is the bottleneck. He is still probably right about the direction.
Noble reads the same week as a warning
George Noble looked at the same facts on Bloomberg and saw the turn he has been waiting for.
But what's happening now is narrative and dreams are meeting lack of cash flow and debt. And that's what the Oracle CDS are telling you. That's what the CoreWeave bonds are telling you.
George Noble, October 22, 2026, Bloomberg Surveillance
He has been short this trade for a while, so discount the conviction. The CDS point holds up, though. Meta's own swaps hit a record the same day.
Does it matter who writes the check?
Yes, and this week showed why. When Meta paid for data centers out of ad profits, the only question was return on capital. Now Meta has negative free cash flow, Tesla has an undrawn $30 billion, the neoclouds are short of working capital and Blackstone is lobbying for an Anthropic listing. The returns on that spending now have to satisfy outside buyers of stock and debt as well as the company. Those buyers get no vote at Meta, and lenders have already started charging more. The demand Gray and Huang describe may well be real. What changed this week is that the companies need outside money to meet it, and that money can be priced, delayed or refused.