AI Funding Shortage Reshapes Industry Narrative: Market No Longer Believes Burning Money Wins

Article body

Jensen Huang had just announced at the GTC Berlin conference that he would join forces with a group of supply chain partners to launch a large-scale industrial chain expansion, covering CPO, substrates, advanced packaging and other segments, and would provide KYEC with a $1.1 billion capacity reservation prepayment. Shortly after, SpaceX released its Q3 financial report: superintelligence business quarterly revenue was about $7.6 billion, corresponding to capital expenditure of as much as $16.4 billion; Meta on the same day raised its full-year capital expenditure plan to $135 billion to $145 billion, and also created a new non-voting Class C stock to pave the way for future large-scale financing.

And in the Class C stock-related documents, Meta explicitly stated that "timing is critical," directly naming Alphabet, SpaceX, Anthropic and OpenAI and saying that other well-known companies have recently launched large-scale financing and more are expected to follow. Therefore, although Meta currently has sufficient capital resources to meet current needs, for future financing needs it needs to raise funds before more companies absorb the available capital in the market.

▌Financing race

But the "AI lacks money" mentioned here does not mean the entire AI industry's funding pool has dried up. At least judging from cash flow estimates of large listed tech companies, the AI investment cycle has not yet reached the point where there is no money to invest—according to a JPMorgan report, global listed tech companies are expected to generate about $2.2 trillion in operating cash flow this year, with capital expenditure and R&D spending totaling about $1.8 trillion, leaving a funding surplus of about $400 billion after offsetting the two. Even if the tech sector raises capital expenditure by 25% per year and cash flow grows by 20% per year from 2026 to 2030, it will still maintain a small surplus by 2030. This means that within a considerable time window, the tech sector as a whole does not need to rely on external capital market funding to support AI infrastructure investment.

But the problem is that a surplus in total funding does not mean every company has money. JPMorgan data shows that Neocloud players account for a relatively small share of the industry, contributing about $30 billion in operating cash flow, corresponding to about $80 billion in capital expenditure and R&D spending; compared with tech giants that have mature businesses and strong balance sheets, such companies are more dependent on external financing. When financing costs rise, funding pressure may quickly transmit to expansion plans.

Coreweave, backed by Nvidia, is one such Neocloud player. Bank of America analyst Tal Liani estimates that CoreWeave needs to raise as much as $20 billion to fulfill customer orders. "No one worries that Meta, Amazon or Google can't borrow money, but if financing thresholds shut small and medium-sized enterprises out, the market doesn't even need a substantive blow-up to fall into panic."

On Thursday local time, Jensen Huang still gave a relatively optimistic statement, saying that market demand for Neocloud computing power is extremely strong, but working capital has become a huge bottleneck. "Because demand in all areas of AI is very strong, capital has become scarce. The scale of funding the AI industry needs far exceeds the financing scale companies can currently obtain, and this problem will eventually be resolved."

As giants successively step up fundraising, funds can no longer flow to every construction project at a sufficiently low cost, and financing costs and financing thresholds rise accordingly. At such times, AI investment is no longer just a competition in technology and efficiency, but also a contest of financial strength and financing capability.

It is worth mentioning that S&P recently also warned that tech giants' AI liabilities may be higher than known levels. Tech giants may support data centers or small AI companies through related-party financing structures (for example, Nvidia and Coreweave), and some related liabilities may not be fully presented in tech giants' public information. "It is reasonable to speculate that there are more (debts) in the market that we have not yet noticed," said Mark Habib, global head of digital infrastructure at S&P Global.

▌Doubts trigger broad selloff; among the Magnificent Seven, only Apple stands out

Once doubt takes root, strong earnings and endorsements from big names may not be enough to reverse market expectations.

Meta's third-quarter revenue and profit both beat market expectations, but the again-raised capital expenditure guidance and the first negative cash flow sent investors' anxiety soaring; the stock then fell nearly 7% in a single day, wiping out about $124 billion in market value in one day.

SpaceX's third-quarter revenue surged 161%, achieving a profit for the first time, with AI business revenue surpassing Starlink for the first time; on the day the financial report was released, the attached stock rose nearly 8%, but the next day it fell more than 9% due to market concerns about AI spending.

Jensen Huang's "Midas touch" also tends to lose effectiveness after repeated high-profile appearances. Several companies Nvidia named at the GTC Berlin conference as co-expanding production either fell sharply after one day of gains or continued their declines.

However, it is worth mentioning that among the battered Magnificent Seven, Apple instead stood out; on Thursday local time it not only rose 1.23% but also hit a record high.

After all, the other "six sisters" bear model R&D/infrastructure costs themselves, while Apple has chosen a more restrained asset-light path: its capital expenditure is only a fraction of other giants', and it uses external models and integrates them into its own product and service ecosystem. This route previously exposed Apple to ridicule for "falling behind in AI," but it also allowed Apple to be rewarded by the market in such turbulent times.

(Source: Cailian Press) Middle of the body Embedded ad