19th-century Parisian Opera crowds were so notoriously hard to please that companies hired paid sycophants, known as claqueurs, to clap, hoot and holler. AI hyperscalers might want to revive the tradition.

Last week, Alphabet reported the biggest quarterly profit in its history, with the $112.1 billion take almost quadrupling what it made a year earlier. The company’s shares fell more than 7%: Investors, racked by anxiety over the company’s massive capital expenditure plans, paid more mind to the Google parent hiking its 2026 spending forecast by $15 billion. Tough crowd. This week, Microsoft, Meta and Amazon get their turns.

Coping With Capex

To find out why higher AI capex has deepened investor worries about hyperscalers’ returns on investment, look no further than the $700 billion in combined capex Alphabet, Microsoft, Meta and Amazon are on track for this year. Wall Street forecasts that figure could top $1 trillion in 2027.

In a research note last week, Moody’s analysts warned capex plans could “threaten credit quality” at Alphabet, Microsoft, Amazon, Meta, Oracle and CoreWeave as they make the costly transition from asset-light models built around software, IP and cloud services to asset-heavy models requiring “unprecedented levels of investment and capital raising.”

“These companies are very large established players with, in many cases, pristine balance sheets, but they’re embarking upon almost unprecedented capital investment,” Moody’s vice president Kevin McNeil said on the ratings firm’s Credit Currents podcast. “And this increase in capital intensity is likely going to pressure credit metrics and is going to lead, in some cases, to negative free cash flow or at least diminished free cash flow.” Moody’s estimates direct debt at the six companies is roughly $460 billion, and McNeil said debt financing “could lead to increasing leverage and additional risk being borne on their balance sheets.” Meanwhile, UBS Global Wealth Management, in its commentary, suggested investors could be alarmed again this week:

• That’s because, according to its analysts, “limited visibility on capex beyond 2027 amid greater investor demand for spending discipline could continue to weigh on investors’ risk appetite, despite solid AI demand.”
• However, UBS advised that, despite the AI capex concerns and inflation worries surrounding the US-Iran conflict, keeping the faith is the best move: “For well-diversified, long-term investors, staying invested remains the most effective strategy to navigate current uncertainty. Given robust earnings growth and an improving cyclical backdrop, we see attractive equity opportunities across sectors and regions.”

Bad Omen: Moody’s emphasized that Alphabet, Amazon, Meta and Microsoft currently have rock-solid balance sheets, meaning their investment grade ratings are in little doubt for now. But Baa2-rated Oracle, to which Moody’s assigned a negative outlook, is only two notches from junk status. Last month, rival ratings firm S&P Global downgraded Oracle to one notch above junk.

There has long been an elephant in the Federal Open Market Committee’s room as it tries to accomplish its mission of promoting maximum employment and stable prices. It’s just much, much bigger now, and it’s hard to ignore a $2 trillion elephant.

That’s roughly how much the US government is issuing in new Treasury bills and bonds each year as the country’s budget deficit swells. Its massive spending problem isn’t new: The US has run a deficit for more than two decades, helped along by the 2008 financial crisis and the COVID-19 pandemic. But the economy is looking relatively healthy now, with inflation down from its decades-high record in 2022, shrinking to 3.5% in June. Weekly jobless claims just dropped to their lowest level since 1969. And yet the government is still issuing increasing amounts of debt. Economists say it’s unsustainable.

Reducing the federal deficit may be Congress’s job, but the central bank, led by Chair Kevin Warsh, has to contend with the economic effects of Treasury borrowing as it determines whether to hike, cut or hold interest rates steady. A committee that already is split on where interest rates should head will gather this week to decide what’s next as a growing chorus of experts point to the problem the federal deficit poses.

Bond Market Blues

The crux of the issue for policymakers lies in the bond market. Investors aren’t going to buy up all the new Treasuries hitting the market just because.

“As the government continues to issue increasing amounts of debt to finance its spending, the supply of Treasury securities in the market grows,” explained Debbie Hippensteel, senior portfolio manager at River Wealth Advisors. “If investor demand does not keep pace with this increased supply, bond prices may decline, resulting in higher yields.”

Those higher yields can then ripple through the economy:

• Borrowing costs for mortgages and auto loans, for instance, go up for both consumers and businesses, which is also what happens when the Fed raises the federal funds rate.
• As a result, even if the Fed doesn’t raise rates, financial conditions may still tighten, which complicates the decision.

FOMC’s July Meeting: The Fed is widely expected to keep interest rates unchanged when it announces its decision Wednesday after the monetary policy committee’s two-day meeting. CME’s FedWatch tool placed the likelihood of the central bank keeping rates steady at 64.2% as of Friday morning. But a hike could be coming soon. FedWatch puts the odds of the Fed raising rates at the September meeting at 80.1%.

OpenAI’s models broke out of a contained testing environment last week to find answers in the wider world, not unlike Stitch in Lilo & Stitch. The publicly available GPT-5.6 Sol and another unreleased model last week cheated on a test by sneaking out of their sandbox and launching a cyberattack on NYC-based developer platform Hugging Face.

The ironic part: The test that OpenAI’s model cheated on was a hacking challenge. And while the attack itself won’t have lasting consequences, containing AI going forward isn’t as easy as stranding it in Hawaii, unlike Stitch.

Double Dilemma

AI companies are simultaneously creating a problem and trying to solve it as their models advance both cybersecurity and cyberattacks. That’s concerning both when rogue actors use the AI maliciously and when, as last week’s incident showed, the AI itself goes rogue.

The potential risks have raised national security red flags as the US’s top AI companies warn Chinese rivals could be copying their most advanced models:

• US Treasury Secretary Scott Bessent said last week the Trump administration is looking into whether Chinese AI companies have been training their models on the US’s most advanced offerings, a process called distillation. The “IP theft,” as Bessent called it, would not only give China a cheaper way to zoom ahead in the AI race but provide hackle-raising hacking skills that rival the models they copy. Experts disagree about whether Chinese AI companies like MoonshotAI depend on distillation to the extent they’ve been accused of.
• But companies including Microsoft, Meta, Nvidia and Hugging Face signed an open letter Friday asking policymakers not to impose restrictions on open-weight AI models, Chinese or otherwise. Hugging Face, which is led by French entrepreneur Clément Delangue, used a powerful Chinese AI model, Z.ai’s GLM 5.2, to defend itself against the OpenAI cyberattack.

In their DNA: Experts say OpenAI’s models didn’t really go “rogue” because AI’s moral compass has been skewed all along. In tests, models have tended to cheat without qualms, with GPT-5.6 Sol trying to cheat in nearly 13% of test runs. Models also tend to lie about cheating after the fact and don’t describe what they did as wrong more than half of the time.