Strong Earnings, Falling Stocks: Who Is Funding AI Infrastructure?

iconTechFlow
Share
AI summary iconSummary
Funding rates are shifting as major tech stocks declined last week despite strong AI earnings. The market is now focused on how AI infrastructure is being funded—through internal cash, prepayments, or debt. Altcoins to watch may emerge as companies compete to fill the funding gap. This trend could determine which companies survive until AI monetization takes off.

Author: Alea Research

Compiled by Deep潮 TechFlow

DeepSight Summary: Last week, nearly all earnings reports confirmed the real demand for AI computing power, yet major tech stocks declined. The market is revaluing companies based on a new criterion: who will fill the funding gap for AI infrastructure—company cash flow, customer prepayments, or debt? This determines which companies can survive until AI becomes profitable.

Growth-oriented companies have been sold off—where is the funding gap, and who is paying for AI infrastructure?

Last week, most companies reporting earnings saw revenue growth, yet their stock prices still fell. The S&P 500 dropped 0.6%, and the Nasdaq fell 2.1%, with nearly every earnings report confirming the real demand for AI computing power.

Building AI capacity means paying for chips, data centers, memory, and power years in advance, with revenue only arriving much later. The market spent a week categorizing companies: who will fill this gap—company cash, customer funds, or borrowed money?

This article is from the stock section of this week's Pulse report.

image

Chart: Performance of Asset Classes Over One Week: Energy Leads, Long-Duration Assets Under Pressure

Source: Alea Research / Substack

Why does the price drop even when there's good news?

SPY, which weights the S&P 500 by company size, allows a few mega-cap tech stocks to dominate the index, rising 0.1% for the week. RSP holds the same 500 stocks but assigns equal weight to each, rising 0.8%. QQQ, concentrated in tech stocks from the Nasdaq-100, fell 1.1%, while the semiconductor-focused basket SOXX dropped 4.3%. The broader market saw buying pressure on Friday, with selling targeted specifically at large AI-themed stocks.

image

Chart: AI Complex Yield Distribution: Semiconductors delivered a 129% return over one year, significantly outperforming the Nasdaq 100 at 22%.

Source: Alea Research / Substack

When interest rates appear harder to cut, these stocks fell the hardest because their prices depend on expected profits years in the future. When rates rise, holding boring alternatives like bonds earns more simply by waiting, so each future profit must be discounted against this more attractive alternative. Investors aren’t exiting the market—if they were, an equal-weighted basket would fall too. They’re shedding stocks whose value lies furthest in the future, just as the future has become more expensive.

Funding gap

A company generates cash from its operations and spends cash on capital expenditures—such as chips, buildings, and grid connections needed for infrastructure. When capital expenditures exceed operating cash flow, the difference must come from somewhere: either new debt, new equity, or the company’s reserves.

All three became more expensive last week. The cost of debt is directly the interest rate. The cost of equity is higher because rising interest rates push stock prices down, forcing companies to sell more shares to raise the same amount of money. Using reserves means forgoing the interest those reserves could have earned, and that interest rises with interest rates.

Interest rates face two major upward pressures. First, attacks on oil tankers have disrupted transportation along critical shipping lanes, pushing Brent crude above $100 per barrel and raising baseline inflation expectations. Second, import prices rose 7.1% year-over-year, driven by higher costs for computers, semiconductors, and industrial machinery—the very hardware fueling the expansion of AI infrastructure. These pressures constrain the Fed’s room to ease monetary policy, with futures markets pricing in a 75% probability of no change and a 25% probability of a rate hike.

image

Chart: Technical pattern of Alphabet after a 10% weekly decline

Source: Alea Research / Substack

Alphabet performed strongly across all businesses this quarter: cloud revenue grew 82% to $24.8 billion, cloud profit margin expanded from 20.7% to 35.6%, and search revenue still rose 17%. Yet the stock fell 7.2% after earnings because Alphabet spent $44.9 billion on capital expenditures while generating only $39.1 billion in operating cash flow, creating a quarterly gap of $5.9 billion. It also informed investors that it expects capital expenditures to reach $195–205 billion in 2026. No one is questioning demand—so the sell-off reflects a repricing of how these expenditures will be financed.

image

Chart: Alphabet compared to Microsoft, which has stronger cash conversion capabilities

Source: Alea Research / Substack

Oracle reported an annual funding deficit, with capital expenditures of $55.7 billion against operating cash flow of $32 billion, creating a gap of $23.7 billion. Its financing plan calls for a combination of debt and equity issuances totaling approximately $40 billion. Despite securing a Pentagon contract worth up to $7 billion, the stock hit its 52-week low, as the company must spend heavily to build data centers before revenue is realized.

image

Chart: Oracle's stock price hits a 52-week low amid ongoing expansion of its capital expenditure plan.

Source: Alea Research / Substack

Tesla delivered a record 480,126 vehicles, but its operating margin compressed to 1.4%. Operating cash flow was insufficient to cover $5.8 billion in quarterly capital expenditures, resulting in negative free cash flow of $1.1 billion.

image

Chart: Tesla's Q2 sales rebound, but free cash flow ends negative

Source: Alea Research / Substack

Companies that receive payment from customers in advance

Companies that maintain their market capitalization have customers or counterparties that contractually commit to absorbing part of the infrastructure costs. First, customer deposits are cash paid upfront for future delivery, directly financing the factory with buyers’ money. Second, take-or-pay agreements require customers to pay a contracted minimum amount regardless of whether they take delivery, turning future demand into a legal right that lenders are willing to finance. Finally, backlogs—accumulated signed but undelivered orders—do not directly provide funding, but they eliminate the risk that would otherwise cause lenders to charge higher fees.

Micron has 16 multi-year take-or-pay agreements covering approximately one-fifth of its DRAM output through 2030, with a minimum contract revenue of about $100 billion and $22 billion in customer deposits. Its customers are funding the construction of its factories.

image

Chart: GE Vernova's absolute values compared to industry peers

Source: Alea Research / Substack

GE Vernova sells turbines and grid equipment, converting power demand into electricity, and booked $24.2 billion in orders during a quarter with $11.1 billion in revenue; its $176 billion backlog is equivalent to approximately four years of already sold work.

Lockheed Martin’s backlog has reached $230 billion, equivalent to about 2.9 years of revenue. Microsoft is a self-funded version of the same escape route: it spent $31.9 billion on capital expenditures, roughly 68% of its $46.7 billion in operating cash flow, leaving $15.8 billion in cash after infrastructure investments. It is building the same infrastructure as Alphabet but without requiring external funding.

image

Chart: Book-to-bill comparison between Lockheed and Northrop

Source: Alea Research / Substack

Two points to note: Pre-sale revenue is not pre-sale profit. Northrop booked $1.84 in new orders for every dollar delivered, yet its margin still slipped from 11.8% to 10.6%, as orders may be arriving faster than the company can profitably fulfill them. Moreover, diversification among these companies is thinner than it appears from a list of stock tickers: most of the incremental order growth in memory, chips, and power equipment traces back to the budgets of the same few data centers. This is why Micron dropped 6.9% on Friday despite its contracts. Holding memory manufacturers, chip baskets, and turbine makers increases the number of stock tickers faster than it increases diversification.

Microsoft and Amazon earnings reports

Microsoft will report earnings on July 29; it must achieve near 40% growth in Azure, another quarter of operating income growth outpacing revenue, and capital expenditures rising above $40 billion per quarter to remain in the self-funded category. The Fed will also announce its interest rate decision on July 29, so volatility is expected.

image

Chart: Microsoft experienced a year of valuation adjustments ahead of its July 29 earnings report.

Source: Alea Research / Substack

Amazon reports earnings on July 30; free cash flow has declined from $25.9 billion a year ago to $1.2 billion. Its own guidance implies an operating margin of 11.2%, down from 13.1% last quarter; if operating income exceeds $22 billion, it would indicate that cloud revenue is filling cash reserves faster than infrastructure is consuming them.

image

Chart: Amazon's path ahead of its July 30 earnings report

Source: Alea Research / Substack

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.