Cisco, Applied Materials, CoreWeave and Simon Property Group Report Earnings the Week of Aug. 10, 2026

Four Reports, One Argument About Where the Money Comes From

Applied Materials sells the machines that etch and layer silicon into computer chips. Simon Property Group collects rent from mall tenants selling shoes and phone cases. This week they report earnings on back-to-back days, and the pairing is the accident that makes the week worth watching^[1][7].

Between them sit Cisco, the networking-gear giant chasing AI orders, and CoreWeave, a company that rents out AI computing power and owes lenders nearly $25 billion^[3][4]. Four businesses, four different customers, one shared question hanging over all of them: is the AI spending boom paying for itself, or is it running on borrowed money that hasn't come due yet?

Nobody disputes the raw numbers. What's argued over is what the numbers mean.

The Number Everyone Agrees On, Read Two Different Ways

Start with the fact both camps cite without blinking. FactSet, the financial data firm that tracks corporate earnings for investors, reported on August 7 that S&P 500 companies were on pace for 50.4% earnings growth in the second quarter, with 88% of companies already reported^[2]. That's the highest growth rate since the second quarter of 2021.

But 2021 is the catch. That quarter looked huge because it was being compared to the depths of the pandemic slump, when profits had cratered. FactSet's 50.4% figure is also "blended," meaning it mixes real, reported results with estimates for the companies still left to report^[2]. It's a real number. It's just not as clean as a headline makes it sound.

Bulls point to specific, concrete demand behind that growth. Cisco booked $1.3 billion in AI-related orders in a single quarter and raised its full-year AI order target above $4 billion^[3]. Applied Materials was expected to report revenue near $9 billion, up about 23% from a year earlier^[1]. That's not a forecast. That's equipment already sold and installed.

Skeptics don't dispute that revenue is real. They argue it's a lagging indicator, and that the borrowing behind it is the number that actually predicts what happens next^[8].

What $159 Billion in Bonds Buys You

Here's the collision at the center of the week: five major tech companies — Amazon, Alphabet, Meta, Microsoft and Oracle — issued $159 billion in bonds in the first five months of 2026 alone^[8]. That's more than those same five companies borrowed in the previous five years combined.

Alphabet's long-term debt more than doubled to $98 billion in six months^[8]. CoreWeave alone is guiding to $650 million to $730 million in interest expense for the second quarter — money it owes lenders before it earns a dollar of profit^[4]. It's also planning $31 billion to $35 billion in capital spending for the year, spending it can only fund by continuing to borrow^[4].

Why borrow instead of just spending cash on hand? Building an AI data center means buying tens of thousands of GPU chips and the buildings to house them, all up front, years before the computing power gets fully rented out and paid off. Bonds let a company spread that cost over time instead of taking the full hit today. The bet only works if the revenue shows up on schedule and the equipment still has value when the bill comes due.

That's exactly what skeptics question. A data center full of GPUs is financed over many years, but the chips inside it may lose value or become outdated in far fewer years than that^[8]. If that happens, the company is still paying off debt for a machine that's already obsolete. Skeptics also point to what they call circular deals: Nvidia holds a stake in CoreWeave, sells CoreWeave the chips it needs, and has agreed to buy back any computing capacity CoreWeave can't rent out^[8]. To a skeptic, that arrangement makes CoreWeave's demand look partly manufactured by its own supplier.

To the builders, the analogy is railroads or the electric grid: heavy borrowing up front is simply what it looks like to build real, physical capacity^[8][9]. CoreWeave says it already holds roughly $99 billion in signed contracts, meaning much of that debt is matched against revenue customers have already committed to pay^[9]. From that seat, calling it a bubble confuses spending a lot of money with spending it recklessly.

A Second Fight, Playing Out in China

Applied Materials carries a dispute of its own, separate from the debt argument: U.S. export controls on chipmaking equipment.

The company has said tighter rules cut roughly $600 million from its fiscal 2026 revenue^[6]. China's share of Applied Materials' business has fallen from near 40% toward the mid-20% range, and China made up 28% of its systems and services revenue in fiscal 2025^[6]. The company says the rules now block it from supplying China's memory-chip makers and its older-generation chip factories^[6].

The case for the controls, made mainly in the U.S., is that some lost revenue is a fair price for slowing down a strategic rival's chipmaking ability^[6]. Asian and semiconductor trade press tell a different version: they argue the rules mostly just push Chinese buyers toward Chinese-made tools, speeding up the very domestic capability the controls were meant to slow^[6]. Applied Materials itself has an interest in the rules staying narrow and predictable, since it wants to keep as much China revenue as Washington will allow^[6].

The Mall That Isn't Betting on Chips at All

Simon Property Group is the outlier in this lineup, and that's exactly why it matters. It doesn't sell chips, networking gear, or cloud computing. It collects rent from shopping malls and outlet centers, which makes its results a read on ordinary household spending instead of AI momentum^[7].

Simon reported first-quarter FFO — funds from operations, the standard profit measure for real estate companies — of $2.91 per diluted share, up 9.0% from a year earlier^[7]. FFO takes net income and adds back depreciation, the accounting charge for a building supposedly losing value over time. Real estate firms use it because a well-maintained mall usually isn't actually losing value the way the depreciation schedule implies, so FFO is considered a truer measure of the cash a REIT is generating.

That distinction matters because of how REITs are taxed. To keep its special tax status, Simon must pay out most of its taxable income to shareholders, so investors watch its cash coverage of the dividend closely^[7]. Simon raised full-year guidance to $13.10–$13.25 per diluted share and lifted its quarterly dividend to $2.25, up 7.1% from a year earlier^[7]. Occupancy stood at 96.0% in March, up slightly from a year before^[7]. Landlords don't fill that much space if shoppers are pulling back.

How the Story Gets Told Depends on Who's Telling It

The way this week gets covered splits along predictable lines, and it's worth naming the pattern rather than picking a side in it.

Business-friendly outlets tend to lead with the order books and treat a strong quarter as proof the AI boom is funding itself, with regulation and export limits framed as the main risk rather than the debt underneath the buildout^[3][5]. Consumer-facing and left-leaning coverage does the opposite: it leads with the $159 billion in bonds and the gap between roughly $400 billion a year in AI spending and an estimated $50–60 billion in AI revenue, treating strong reported profits as something that hasn't caught up with reality yet rather than a rebuttal to the debt concern^[8].

One piece worth flagging directly: a Moneywise headline reads "Forget the AI bubble. The IMF says the real threat is the mountain of debt behind it — and 60% of planned data centers haven't even broken ground." The debt figures trace to the IMF, but the 60% data-center statistic actually comes from a separate J.P. Morgan analysis, folded under one IMF-attributed headline^[8]. Both numbers check out. The framing that glues them together as a single source doesn't.

Financial media as a whole has its own incentive worth noting: earnings week is a traffic event, and previews get written to make each report feel like a verdict. Watch for the word "test," as in whether the growth "can continue." It frames four unrelated companies — a chipmaking tool seller, a networking company, an AI landlord, and a mall REIT — as one referendum on the economy. A mall's occupancy rate and a GPU company's interest bill are answering different questions. Neither one settles the other, and this week won't either.